DESCRIPTION Title of Invention: ANGLE ERROR CORRECTION DEVICE AND ANGLE ERROR CORRECTION METHOD FOR POSITION DETECTOR, ELEVATOR CONTROL DEVICE, AND ELEVATOR SYSTEM Technical Field [0 001] This invention relates to an angle error correction device and an angle error correction method for a position detector, which are applied to a control device for an elevator hoisting machine, a control device for an in~vehicle motor, a control device for a motor of a machine tool, or the like, for example, in order to correct an angle error in a position detector having a periodic error that is determined univocally in accordance with a rotation position of a motor, and also to an elevator control device and an elevator system. Background Art [0002] In a conventional resolver angle detection device, an angle detector detects an angle signal from a signal detected in a resolver, whereupon an angle error estimator calculates and corrects an angle error using the fact that an error waveform of the resolver is constituted by an n-th order component that is unique to the resolver and reproducible (see PTL 1, for example). [0003] In the resolver angle detection device of PTL 1, a position error is calculated by referring to the detected angle signal, a speed error signal is calculated by differentiating the 1 position error, and a detected error is calculated for each frequency component by performing frequency analysis on the speed error signal using the Fourier transform, for example. Further, an estimated angle error signal is generated by combining the calculated angle errors, and the detected angle signal is corrected by an angle signal correction circuit using the generated estimated angle error signal. Citation List Patent Literature [0004] [PTL 1] Japanese Patent Application Publication No. 2012-145371 Summary of Invention Technical Problem [0005] However, the prior art includes the following problem. In the conventional resolver angle detection device, a motor rotation speed is detected by a speed detector from the angle signal detected by the angle detector, and the angle error is estimated using the detected speed. Here, when the angle error is estimated using the detected speed, the estimation precision of the angle error is determined according to a speed resolution of the angle detector or the speed detector. When the angle detector or the speed detector has a low speed resolution, therefore, a quantization error occurs, and as a result, the angle error cannot be estimated with a sufficient degree of precision. [0006] This invention has been designed to solve the problem described above, and an object thereof is to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately. Solution to Problem [0007] An angle error correction device for a position detector according to this invention is used in combination with a motor control device that controls a motor, a position detector that outputs a position detection signal obtained by detecting a rotation position of the motor, and a current detector that detects a current flowing through the motor to correct a periodic angle error that is included in the position detection signal and determined in accordance with the rotation position, the angle error correction device including: a frequency analysis unit that performs frequency analysis on a detected current detected by the current detector while rotating the motor, calculates an amplitude and a phase of a specific frequency, and outputs the amplitude and the phase of the specific frequency as a frequency analysis result; an angle error correction unit that outputs an addition signal, the addition signal being obtained by adding the position detection signal to an input signal input therein, to the motor control device; and an angle error estimation unit that implements first control processing and second control processing repeatedly on a plurality of different test signals, the first control processing being processing for inputting a set value of a test signal having known amplitude, phase, and frequency values into the angle error correction unit as the input signal and having the angle error correction unit operate the motor by applying a test signal corresponding to the set value thereto, the second control processing being processing lor having the frequency analysis unil perform frequency analysis at the frequency of the test signal on the detected current obtained in the first control processing, estimates estimated values of an amplitude and a phase of the angle error on the basis of amplitudes and phases constituting at least two types of frequency analysis results calculated by the frequency analysis unit in the second control processing, and outputs the estimated values to the error correction unit, wherein the angle error correction unit outputs the addition signal to the motor control device using the estimated values of the amplitude and the phase of the angle error as the input signal. [0008] Further, an angle error correction method for a position detector according to this invention is executed by an angle error correction device for a position detector, the angle error correction device being used in combination with a motor control device that controls a motor, a position detector that outputs a position detection signal obtained by detecting a rotation position of the motor, and a current detector that detects a current flowing through the motor to correct a periodic angle error that is included in the position detection signal and determined in accordance with the rotation position, the angle error correction method including: a frequency analysis step in which frequency analysis is performed on a detected current detected by the current detector while rotating the motor, an amplitude and a phase of a specific frequency are calculated, and the amplitude and the phase of the specific frequency are output as a frequency analysis resulL; an angle error correction step in which an addition signal obtained by adding the position detection signal to an input signal input therein is output to the motor control device; a first control step in which a set value of a test signal having known test amplitude, phase, and frequency values is input as the input signal during the angle error correction step, and the motor is operated by applying a test signal corresponding to the set value thereto; a second control step in which frequency analysis is performed during the frequency analysis step at the frequency of the test signal on the detected current obtained in the first control step; the first control step and the second control step being executed repeatedly on a plurality of different test signals, a third control step in which estimated values of an amplitude and a phase of the angle error are estimated on the basis of amplitudes and phases constituting at least two types of frequency analysis results calculated in the frequency analysis step during the second control step; and a fourth control step in which the addition signal is output to the motor control device using the estimated values of the amplitude and the phase of the angle error as the input signal during the angle error correction step. [0009] Further, an elevator control device according to this invention includes a motor control device that controls a hoisting machine of an elevator, a position detector that detects a rotation position of the hoisting machine, and includes a periodic error determined univocally in accordance with the rotation position, a current detector that detects a current flowing through the hoisting machine, and an angle error correction device that is connected to the motor control device, the position detector, and the current detector. Advantageous Effects of Invention [0010] With the angle error correction device and angle error correction method for a position detector according to this invention, the position detector can perform frequency analysis on a specific frequency component of the motor current, and estimate the periodic error of the position detector on the basis of the frequency analysis result. At this time, processing for performing an operation by applying the test signal having known amplitude, phase, and frequency values and performing frequency analysis at the frequency of the test signal is implemented a plurality of times, whereupon the error of the position detector is estimated on the basis of amplitudes and phases calculated in the plurality of frequency analyses. As a result, it is possible to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately. [0011] Further, with the elevator control device according to this invention, the position detector can perform frequency analysis on a specific frequency component of the motor current, and estimate the periodic error of the position detector on the basis of the frequency analysis result. At this time, processing for performing an operation by applying a test signal having known amplitude, phase, and frequency values and performing frequency analysis at the frequency of the test signal is implemented a plurality of times, whereupon the error of the position detector is estimated on the basis of amplitudes and phases calculated in the plurality of frequency analyses. As a result, it is possible to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately in an elevator system. Brief Description of Drawings [0012] Fig. lis a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a first embodiment of this invention. Fig. 2 is a block diagram showing a configuration of a motor control device of the motor control system shown in Fig. 1. P^ig. 3 is an illustrative view showing a relationship between a motor rotation angle and a detected angle including a periodic error in a position detector to which the angle error correction n device for a position detector according to the first embodiment of this invention is applied. Fig. 4 is a block diagram illustrating an output of an angle error correction unit in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 5 is a block diagram illustrating a transfer characteristic from the output of the angle error correction unit to a motor current in the form of a transfer function expression in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 6 is a Bode diagram showing an example of the transfer characteristic from the output of the angle error correction unit to the motor current in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 7 is a flowchart showing processing executed by an angle error estimation unit of the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 8 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a second embodiment of this invention. Fig. 9 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a third embodiment of this invention. Fig. 10 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a fourth embodiment of this invention. Fig. 11 is a flowchart showing in detail processing for calculating an estimated angle error value, which is executed by the angle error estimation unit of the angle error correction device for a position detector according to the fourth embodiment of this invention. Fig. 12 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a fifth embodiment of this invention. Fig. 13 is a view showing a configuration of an elevator control device according to a sixth embodiment of this invention. Description of Embodiments [0013] Preferred embodiments of an angle error correction device and an angle error correction method for a position detector according to this invention will be described below using the drawings. Note that identical or corresponding parts of the drawings will be described using identical reference numerals. [0014] First Embodiment Fig. 1 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a first embodiment of this invention. In Fig. 1, the motor control system includes a motor control device 1, a motor 2, a position detector 3, a current detector 4, and an angle error correction device 5. [0010] The motor control device 1 is a device for controlling a rotation speed or a rotation position of the motor 2. Here, referring to Fig. 2, a case in which the motor control device 1 controls the rotation speed of the motor 2 will be described. Fig. 2 is a block diagram showing a configuration of the motor control device provided in the motor control system shown in Fig. 1. [0016] In Fig. 2, the motor control device 1 includes a speed calculation unit 11, a speed control unit 12, a current control unit 13, and a power converter 14. The speed calculation unit 11 calculates the rotation speed of the motor 2 on the basis of position information (a corrected rotation position of the motor 2) or angle information obtained when the angle error correction device 5 corrects the rotation position (a motor rotation position) of the motor 2, detected by the position detector 3. Note that the speed calculation unit 11 calculates the rotation speed most simply by time-differentiating the position information or the angle information. [0017] The speed control unit 12 calculates a current command value (a torque command value) for the motor 2 so that the rotation speed of the motor 2 reaches a desired speed. The current control unit 13 calculates a voltage command value for the motor 2 so that a current (a motor current) of the motor 2, detected by the current detector 4, matches the current command value output by the speed control unit 12. The power converter 14 applies an applied motor voltage to the motor 2 on the basis of the voltage command value output by the current control unit 13 in order to control the motor current. [0018] Here, P control, PI control, and PID control are typically used as the control executed by the speed control unit 12 and the current control unit 13, but various other control methods may be used. Further, vector control is typically used to control the current of the motor 2, in which case the motor current and the applied motor voltage are converted to d-q axes, whereupon a control method such as the aforesaid PID control is applied to the converted current and voltage. [0019] Furthermore, an inverter is typically used as the power converter 14 that applies a voltage to the motor 2. The inverter converts a voltage from a power supply, not shown in the drawings, into a desired variable-voltage, variable-frequency voltage. In this invention, a power converter that converts an AC voltage into a DC voltage using a converter and then converts the DC voltage into an AC voltage using an inverter, such as a typical, commercially available inverter device, or a variable-voltage, variable-frequency power converter including a power converter that converts an AC voltage directly into a variable-voltage, variable-frequency AC voltage, such as a matrix converter, may be used. [0020] Further, the inverter according to the first embodiment of this invention may include a coordinate conversion function in addition to the inverter function described above. More specifically, when the voltage command value is a voltage command value on d-q axes, the inverter is assumed to include a coordinate conversion function for converLing the vollage command vdlue on the d-q axes into a voltage corresponding to the instructed voltage command value by converting the voltage command value on the d-q axes into a phase voltage or a line voltage. Note that this invention may be applied likewise when a device or means for correcting a dead time of the inverter is provided. [0021] Moreover, when the rotation position of the motor 2 is controlled, a position control unit is added on a higher order than the speed control unit 12. The position control unit calculates a speed command value for the motor 2, which is input into the speed control unit 12 so that the rotation position of the motor 2 reaches a desired position. The speed control unit 12 executes the control described above using the speed command value as a desired speed. P control, PI control, and PID control are typically used as the control executed by the position control unit, but various other control methods may be used. [0022] Alternatively, the speed control unit 12 may be omitted such that the motor control device is constituted by the position control unit and the current control unit 13. In this case, the position control unit calculates the current command value for the motor 2 so that the rotation position of the motor 2 reaches the -i o desired position. [0023] Returning to Fig. 1, the current detector 4 measures a current of the motor 2. When the motor 2 is a three-phase motor, for example, a two-phase phase current is typically measured. However, a three-phase phase current may be measured instead. [0024] Further, the position detector 3, which is constituted by an optical encoder, a magnetic encoder, or a resolver, for example, detects the rotation position of the motor 2, which is required to control the motor 2, and outputs a position detection signal. Furthermore, as shown in Fig. 3, rotation position information output by the position detector 3 includes a periodic error determined univocally in accordance with the rotation position of the motor 2. [0025] Here, a periodic error determined univocally in accordance with the rotation position of the motor 2 is an error that can be reproduced in accordance with the rotation position, such as the detection error of the resolver described in paragraphs 0020 and 0021 of PTL 1, or a missing pulse or an imbalance in an inter-pulse distance caused by a slit defect in an optical encoder, for example. [0026] Hereafter, the periodic error determined univocally in accordance with the rotation position of the motor 2 will be expressed as an angle error 0err that is obtained by converting the position information into an angle. Note that this invention may be applied to a case in which the position detector 3 includes a periodic error determined univocally in accordance with the rotation position of the motor 2, and a main component order of the angle error Gerr is known. [0027] The periodic angle error 0crr of the position detector 3 can be expressed approximately using a sine wave, as shown below in Equation (1) . Note that since there is essentially no difference between sine wave notation and cosine wave notation, sine wave notation is used throughout the first embodiment of this invention. [0028] [Math.1] 0err * 4 sm(NxOm + (Px) + A2 sm(N2&m + „)... ( 1) [0029] In Equation (1), 0m denotes a mechanical angle of the motor 2, A! denotes an error amplitude of an Nx-th order, A2 denotes an error amplitude of an N2-th order, An denotes an error amplitude of an Nn~th order, §i denotes a phase shift (an initial phase) of the Ni-th order relative to the mechanical angle of the motor 2, 2 denotes a phase shift of the N2-th order relative to the mechanical angle of the motor 2, and (|>n denotes a phase shift of the Nn-th order relative to the mechanical angle of the motor 2. [0030] Note that the spatial orders Ni, N2, . . . , Nn in Equation (1) are spatial orders of a main component of the periodic error determined univocally in accordance with the rotation position of the motor 2, and need not be consecutive integers such as 1, 2, . . . , Nn. Here, the main component is a component in which an amplitude of the component at the corresponding spatial order is larger than amplitudes at other frequencies. 14 [0031] Further, in Equation (1) , three or more frequency components are combined, but the frequency components of the periodic angle error 0err may be constituted by one, two, or more components . [0032] In the first embodiment of this invention, the angle error correction device 5 includes a frequency analysis unit 51, an angle error estimation unit 52, and an angle error correction unit 53. The angle error estimation unit 52 estimates the periodic angle error of the position detector 3 as a function of Equation (1) on the basis of a frequency analysis result obtained when the frequency analysis unit 51 analyzes the motor current. [0033] Further, the angle error correction unit 53 generates an addition signal by adding the position detection signal from the position detector 3 to an input signal from the angle error estimation unit 52. More specifically, the angle error correction unit 53 generates an angle error correction value in an identical format to Equation (1) on the basis of an angle error estimation result, and corrects the angle error by adding the generated angle error correction value to the information indicating the rotation position or the angle of the motor 2. [003 4] An operation of the angle error correction device 5 will now be described. The frequency analysis unit 51 calculates at least one of the amplitude and the phase of the motor current at a specific frequency on the basis of the motor current detected by the current detector 4 and the information indicating the rotation position or the angle of the motor 2, i.e. the output of the position detector 3. [0035] Note that although the information indicating the rotation position or the angle of the motor 2 is input into the frequency analysis unit 51 in Fig. 1, the invention is not limited to this configuration, and position information or angle information obtained when the angle error correction unit 53 corrects the rotation position of the motor 2 may be input instead. [0036] Here, the frequency analysis unit 51 is preferably configured to obtain the amplitude and phase at a desired frequency of the input signal by means of Fourier transform, Fourier series analysis, or fast Fourier transform, but may be configured to extract a desired frequency signal by means of a filter combining a notch filter and a band pass filter, and then calculate the desired amplitude and phase of the input signal using an amplitude detection unit and a phase detection unit. Further, the filter used in this case may be an electric filter combining a resistor, a capacitor, a coil, and so on, or may take the form of processing executed by a computer. [0037] Furthermore, the motor current input into the frequency analysis unit 51 may be any current used in vector control, such as a d axis current, a q axis current, a y axis current, a 5 axis current, an a axis current, or a (3 axis current, these currents being obtained by subjecting a phase current to coordinate conversion. [0038] Note that here, a signal having a desired frequency (a specific frequency) denotes a signal that is derived from the periodic angle error 0err of the position detector 3 and has an identical frequency to the main component of the angle error 9err/ or a signal having an identical frequency to the main component of a test signal generated by the angle error correction unit 53. Note that the test signal will be described in detail below. Further, in the first embodiment of this invention, the desired frequency is expressed as a spatial frequency, but a temporal frequency is essentially no different thereto. [0039] Here, the spatial frequency denotes a frequency within a specific interval, which in the first embodiment of this invention corresponds to a single rotation of the motor 2 . Moreover, a signal including N periodic waves within a single mechanical angle rotation of the motor 2 will be referred to as a wave of the spatial order N. [0040] In the control device for the motor 2 having the position detector 3, the error of the position detector 3 occurs periodically in accordance with the rotation position of the motor 2, and therefore frequency analysis preferably involves analysis by means of spatial frequency. Likewise in Equation (1) , the angle error Gerr is expressed in terms of the spatial frequency, and moreover, the inputs of the frequency analysis unit 51 shown in Fig. 1 are inputs (a current and an angle) corresponding to spatial frequency analysis. [0041] However, the first embodiment of this invention can also be applied to frequency analysis by means of temporal frequency, and when frequency analysis is performed by means of temporal frequency, the frequency analysis is performed using the detected angle, a measured time measured by a time measurement unit, and the current as inputs in place of the current and the angle. [0042] The angle error estimation unit 52 estimates the angle error 9err determined univocally in accordance with the rotation position of the motor 2 by means of an estimation method to be described below using a current amplitude value, or a current amplitude value and a phase value, of the desired frequency component, these values serving as the output of the frequency analysis unit 51, and either the rotation position of the motor 2, which serves as the output of the position detector 3, or information indicating the angle corrected by the angle error correction unit 53 . The angle error estimation unit 52 then outputs the estimated angle error value to the angle error correction unit 53 as angle information or position information. [0043] The angle error correction unit 53 adds an angle error correction signal based on the estimated angle error value serving as the output of the angle error estimation unit 52 to the rotation position of the motor 2, which serves as the output of the position detector 3, and outputs corrected position information or angle information. [0044] A case in which the estimated angle error value is output in an identical unit system to the output signal of the position detector 3 will now be described as an example. When the position detector 3 is an optical encoder having a resolution of 1024 pulses per rotation, and the estimation result obtained by the angle error estimation unit 52 is 1°, the angle error estimation unit 52 outputs three pulses, which is the number of pulses corresponding to 1°, as the position information. [0045] As shown in Equation (1) , when the angle error includes a plurality of frequency components, the angle errors of the respective components may be estimated in succession and then added together, or the plurality of frequency components may be estimated simultaneously. By employing simultaneous estimation at this time, an estimation time can be shortened in comparison with a case where the angle errors of the respective components are estimated in succession. Here, for simplicity, a case in which the angle error includes only one frequency component will be described. [0046] It is known that when speed feedback control or position feedback control is implemented by the position detector 3 that includes a periodic angle error determined univocally in accordance with the rotation position of the motor 2, current pulsation or current command value (torque command value) pulsation that includes a frequency component of the same order as the angle error is generated. [0047] Similarly, it is known that when speed feedback control or position feedback control is implemented by having the angle error correction unit 53 generate a specific periodic signal and adding the generated specific period signal to the output of the position detector 3, current pulsation or current command value (torque command value) pulsation that includes an identical frequency component to the added signal is generated. [0048] When, dL Lhis Lime, the frequency analysis unit 51 performs frequency analysis on the phase current in a case where the motor 2 is a permanent magnet synchronous motor, the spatial order of the current pulsation appearing in the phase current is a Pn ± Nn-th order, where Pn denotes a number of pole pairs and Nn denotes the order of the desired frequency. [0049] It is therefore sufficient to perform frequency analysis on the current of at least one phase of the phase current, and estimate a Pn + Nn-th order or Pn - Nn-th order angle error from a Pn + Nn-th order or Pn - Nn-th order current. Note, however, that when the order Nn of the desired frequency is larger than the number of pole pairs Pn of the motor 2, the Pn - Nn-th order may take a negative value so as to become non-existent, and therefore, in this case, frequency analysis is performed on the Pn + Nn~th order current. Moreover, a constant-torque, constant-speed operation is preferably implemented when estimating the angle error. [0050] Further, when the frequency analysis unit 51 performs frequency analysis on either a d axis current or a q axis current and the order of the desired frequency is Nn, current pulsation components appearing on the dq axes have pulsation components of an identical order to the Nn-th order. Furthermore, the q axis current, which is a torque current, flows around the d axis current due to a magnetic pole deviation caused by position pulsation of the desired frequency, and therefore the d axis current exhibits current pulsation analogous to the angle error. Moreover, speed pulsation in Ihe q axis currenL forms current coruruand value (a torque command value) pulsation via a speed control system. Accordingly, the q axis current forms current pulsation analogous to the angle error that causes the speed pulsation. [0051] Note that when frequency analysis is performed using a current detection value or a current command value fa torque command value) of the d axis current or the q axis current, estimation is performed in a condition where the q axis current to be fed back is fixed, or in other words in a condition of constant acceleration. In particular, estimation is preferably performed in a condition where the acceleration is zero, or in other words a condition in which the motor 2 rotates at a constant speed. [0052] The angle error and the signal from the angle error correction unit 53 generate current pulsation in accordance with a transfer characteristic determined according to dynamic characteristics of the motor control device 1, the motor 2, and a load connected to the motor 2. Hence, when the transfer characteristic can be determined, it is possible to estimate the angle error signal by which the current pulsation is generated. In other words, the angle error by which the current pulsation is generated can be determined by back calculation from the determined transfer characteristic and the current pulsation. [0053] A method of estimating the transfer characteristic and the periodic angle error component from a frequency analysis result obtained in relation to a current pulsation component will now be described. Note that frequency analysis may be performed on a current command value (torque command value} pulsation component instead of the current pulsation component, but a case in which frequency analysis is performed on the current pulsation component will be described below. [0054] Fig. 4 is a block diagram illustrating the output of the angle error correction unit in the angle error correction device for a position detector according to the first embodiment of this invention. In Fig. 4, the angle error correction unit 53 generates an angle error correction signal. The corrected rotation position of the motor 2, which is corrected by adding the angle error correction signal generated by the angle error correction unit 53 to the motor rotation position serving as the output of the position detector 3, is fed back to the motor control device 1. The motor current detected by the current detector 4 is also fed back to the motor control device 1. [0055] When, at this time, a transfer characteristic from the angle error correction signal to the motor current is represented in the form of a transfer function expression as Gerr i (s) , a block diagram thereof is as shown in Fig. 5. Fig. 5 is a block diagram illustrating the transfer characteristic from the output of the angle error correction unit to the motor current in the form of a transfer function expression in the angle error correction device for a position detector according to the first embodiment of this invention. [0056] Here, "s" is a Laplace operator. Further, Gerr_i (s) also matches a transfer function from the motor rotation position, which serves as the output of the position detector 3, to the motor current. Note that when a load is connected to the motor 2, the dynamic characteristic of the load is also represented by Gerr_i (s) . [0057] In this invention, a gain and a phase of Gerr_i (s) at the frequency of the angle error or a specific frequency are determined, whereupon the angle error is estimated from the determined gain and phase. Fig. 6 shows an example of Gerr_i (s). Fig. 6 is a Bode diagram showing an example of the transfer characteristic from the output of the angle error correction unit to the motor current in the angle error correction device for a position detector according to the first embodiment of this invention. [0058] In Fig. 6, an upper section shows a gain characteristic and a lower section shows a phase characteristic. When the angle error frequency varies, the amplitude and the phase of motor current pulsation corresponding to the angle error vary in accordance with the characteristics shown in Fig. 6. Further, the angle error frequency varies according to the rotation speed of the motor 2. In other words, the phase and the amplitude of current pulsation derived from the angle error vary in accordance with the rotation speed. [0059] Next, processing executed by the angle error estimation unit 52 according to the first embodiment of this invention will be described with reference to a flowchart shown in Fig. 7. [0060] First, at the start of angle error estimation, the angle error estimation unit 52 outputs an operation command for operating Lhe motor 2 Lo Lhe motor control device 1, cuid outputs a command to set the angle error correction signal at zero, or in other words a command to set the test signal at zero, to the angle error correction unit 53 (step SI). As a result, the angle error correction unit 53 sets the angle error correction signal at zero such that the motor 2 is rotated in a condition where angle error correction is not performed. [00 61] Next, the angle error estimation unit 52 outputs a frequency analysis command to the frequency analysis unit 51 to cause the frequency analysis unit 51 to perform frequency analysis on the motor current (step S2). The frequency analysis result is input into the angle error estimation unit 52. At this time, frequency analysis is performed on the motor current at a frequency corresponding to the angle error frequency. [0062] This is achieved by determining Fourier coefficients of the current pulsation corresponding to the specific frequency, for example. A case in which the Fourier coefficients of the current pulsation are determined with respect to the q axis current will be described below. The Fourier coefficients at a frequency Mi [Hz] of the current pulsation of the q axis current, which corresponds to the angle error frequency, can be determined by solving arithmetic expressions in Equations (2) and (3), shown below. [0063] [Math.2] r 2 n , , Ki = ™\ iq(t)cos(2nM1t) dt • • • (2) T J_z 2 [0064j [Math.JJ r 2 fz , , Sm=^ ig(t)sin(27rM1t) dt • * * (3) [0065] In Equations (2) and (3), iq (t) denotes the value of the q axis current, and T denotes a current pulsation period of the frequency Mi [Hz] . Note that T = 1 /Mi. Further, Anl and Bni denote a cosine wave coefficient and a sine wave coefficient, respectively. [0066] Equations (2) and (3) show a case in which the Fourier coefficients are determined by being integrated over time, but the Fourier coefficients may be determined by being integrated in accordance with the rotation angle of the motor 2. Further, Equations (2) and (3) are arithmetic expressions in a continuous time region, but when Equations (2) and (3) are packaged in a computer such as a microcomputer, the equations are packaged after being converted into expressions in discrete time regions. Furthermore, Equations (2) and (3) can be solved using cosine wave and sine wave signal generators, a multiplier, and an integrator, and can therefore be packaged in a computer easily. [0067] Moreover, in Equations (2) and (3), the Fourier coefficients are calculated by being integrated over a single signal period, but may be determined as a value that is integrated over a number of periods and then divided by the number of periods. In this case, the Fourier coefficients are determined as an average value of the number of periods, and therefore the effects of irregularities and disturbances on the current pulsation can be reduced. Furthermore, an inLeyraLion start Lime preferably slarls from a reference point (zero degrees, for example) of the rotation angle of the motor 2. As a result, the Fourier coefficients can be determined using the rotation angle of the motor 2 as a reference. [0068] Here, an amplitude An and a phase (j)n of the current pulsation component can be determined from the Fourier coefficients of Equations (2) and (3) using Equations (4) and (5), shown below. [0069] [Math.4] An = ^A2nl + Bl± • - ■ (4) [0070] [Math.5] n determined in Equations (4) and (5) . Note that instead, the Fourier coefficients Ani and Bni may be stored, and the amplitude and phase may be determined by solving Equations (4) and (5) . [0072] Next, the angle error estimation unit 52 operates the motor 2 by applying the test signal thereto (step S3) such that frequency analysis is performed on the motor current in a condition where the motor 2 is rotated (step S4) . The rotation speed of the motor 2 at this time is set at an identical speed to step SI. Note that an operation command for operating the motor 2 and a set value of the test signal are output by the angle error estimation unit 52. Further, the frequency analysis result is input into the angle error estimation unit 52. [0073] Here, the test signal is a sine wave or cosine wave test signal set at an amplitude, a frequency, and an initial phase that have been determined in advance. The test signal is generated by the angle error correction unit 53 and added to the output of the position detector 3. A sine wave and a cosine wave can be converted into each other by varying the respective initial phases thereof, and therefore a sine wave will be used in the following description. Furthermore, the predetermined amplitude of the test signal is set at At, and the initial phase thereof is set at (|>t. [0074] In the first embodiment of this invention, a sine wave signal having a different frequency to the angle error frequency is applied as the test signal. For example, a sine wave signal having a frequency in the vicinity of the angle error frequency is used as the test signal. Note that the vicinity denotes frequencies that are larger or smaller than the angle error frequency by approximately 10% to 20%, for example, so that the gain and the phase at the frequency of the test signal are within a range that may be considered substantially equal to the gain and the phase of the angle error frequency. [0075] By setting the frequency of the test signal at a different value to the angle error frequency in this manner, the test signal and the angle error correction signal do not intersect, and therefore the frequency analysis can be performed easily, thereby facilitating calculation of the transfer characteristic and estirnation of Ihe angly exioi. [0076] Further, when frequency analysis is performed on the motor current in step S4, the frequency analysis is performed at the current pulsation frequency corresponding to the test signal applied in step S3. Note that when the frequency analysis is performed on the d axis current or the q axis current, an identical frequency to the frequency of the test signal is obtained. The frequency analysis is performed using similar calculations to those shown in Equations (2) to (5) , but the values of Mi and T are replaced with a frequency and a period corresponding to the test signal. At this time, the amplitude and the phase of the current pulsation, determined from the Fourier coefficients of the current pulsation, are set respectively at Ait and §n. [0077] Next, the angle error estimation unit 52 calculates the estimated angle error value of the position detector 3 (step S5}. In step S4, the current pulsation amplitude corresponding to the amplitude At of the test signal is Alt, and it is therefore evident that the current pulsation is obtained by multiplying the amplitude of the test signal by hn/At. [0078] Here, when the angle error frequency and the frequency of the test signal are close, a magnification ratio of the current pulsation amplitude of the current pulsation generated by the angle error relative to the angle error amplitude may be considered identical to the aforesaid magnification ratio, and as a result, the amplitude of an error signal at which the current pulsation amplitude An is yenerctted can be determined in relation to the error signal determined in step S2. In other words, when the amplitude of the error signal is set at AL, the amplitude can be determined from Equation (6), shown below. [0079] [Math.6] Ai=^At ■ • • (6) [0080] Further, in step S4, the current pulsation phase corresponding to the initial phase t of the test signal is it, and it is therefore evident that the current pulsation phase deviates from the phase of the test signal by ii is generated can be determined in relation to the error signal determined in step S2 . In other words, when the phase of the error signal is set at <|>i, the phase can be determined from Equation (7), shown below. [0082] [Math.7] .it of the first embodiment are determined using the Fourier coefficient of the current pulsation component derived from the extracted test signal, whereupon the estimated angle error value is determined from Equations (6) and (7) using An and (j)j_i determined in step S2. Note that At and t in Equations (6) and (7) respectively denote the amplitude and the initial phase of the test signal, and similarly to the first embodiment are known. [0118] in the third embodiment of this invention, as described above, the motor 2 is operated in a condition where the frequency of the test signal matches the angle error frequency, and the angle error is estimated from the corresponding current pulsation component. At this time, the transfer characteristic determined during the operation performed by applying the test signal can be aligned precisely with the angle error frequency, and therefore the angle error can be estimated with an even higher degree of precision. Moreover, in contrast to the second embodiment, the motor speed remains the same during the operation performed without applying the test signal and the operation performed by applying the test signal, and therefore the cause of an estimation error in the angle error due to a difference in the operating speed can be eliminated. [0119] Further, the angle error may be estimated a plurality of times using the procedures described above with respect to a plurality of test signals having different initial phases and amplitude values, and an average value thereof may be set as the estimated angle error value. Moreover, this embodiment can be expanded easily to a case in which the angle error includes a plurality of frequency components. [012 0] FourLh Embodiment In a fourth embodiment of this invention, the operations of the angle error estimation unit 52 differ from the first embodiment. Processing executed by the angle error estimation unit 52 according to the fourth embodiment of this invention will be described below with reference to a flowchart shown in Fig. 10. [0121] Note that in Fig. 10, flows marked with identical reference symbols to Fig. 7 denote identical operations to the first embodiment, and description thereof has been omitted. In the fourth embodiment of this invention, operations performed in step S31 and step S33, in which test operations are performed by applying test signals, and step S35, in which the estimated angle error value is calculated, differ from the first embodiment. [0122] In the fourth embodiment of this invention, an operation is performed by generating two test signals in which at least one of the amplitude and the initial phase is different in each test signal, whereupon the angle error is estimated using frequency analysis results obtained in relation to current pulsation components corresponding to the test signals. The frequencies of the test signals are set to be identical to the angle error frequency. [0123] First, in step S31, the motor 2 is operated by applying a first test signal having a predetermined amplitude, a predetermined initial phase, and an identical frequency to the angle error frequency. Next, in step S2, an identical operation to the first embodiment is performed. [0124] Next, in step S33, a similar operation to step 531 is performed, except that a second test signal is set to differ from the first test signal applied in step S31 in terms of at least one of the amplitude and the initial phase of the applied test signal. Next, in step S4, an identical operation to the first embodiment is performed. [0125] Next, in step S35, the angle error is estimated using frequency analysis results obtained during the two test operations implemented in the preceding steps. Processing executed by the angle error estimation unit 52 in step S35 of Fig. 10 to calculate the estimated angle error value will now be described in detail with reference to a flowchart shown in Fig. 11. [0126] Note that the frequency analysis results determined in step S2 and step S4 are frequency analysis results obtained in relation to current pulsation generated in response to a combined signal combining the angle error and the test signal. First, a calculation is performed to separate the current pulsation component derived from the test signal from these frequency analysis results (step S41) . [0127] As an example of this calculation, the current pulsation component derived from the test signal can be separated by subtracting the frequency analysis result obtained in step S2 from the frequency analysis result obtained in step S4, for example. In other words, the difference between the Fourier coefficients determined in step S4 and step S2 serves as the Fourier coefficient of the current pulsation component derived from the test signal. [0128] Huwevei, alLenLion must be paid Lo Lhe fact. LhuL the test signal at this time is a signal (for convenience, referred to as a combined test signal) obtained by performing subtraction on the test signals of step S2 and step S4. In other words, the two test signals are known, and therefore the combined test signals obtained by performing subtraction on the test signals are also known, meaning that the amplitudes and the initial phases thereof are known. [0129] Next, the transfer characteristic at the frequency of the angle error (which is equal to the frequency of the combined test signal) is determined from a relationship between the current pulsation component derived from the combined test signal extracted in step S41 and the combined test signal by implementing similar procedures to the first embodiment (step S42) . As a result, an amplitude and a phase corresponding to Ait and (j>it of the first embodiment can be determined using the extracted combined test signal and the Fourier coefficient of the current pulsation component derived from the combined test signal. [0130] Next, the current pulsation component derived from the angle error is extracted from the frequency analysis result obtained during the test operation (step S43). Here, the test operation may be either the test operation of step S31 or the test operation of step S33. A case in which the test operation result of step S31 is used will be described below. [0131] The first test signal used during the test operation of step 331 is known, and therefore the amp 11 Lu.de arid Lhe phase of the current pulsation derived from the first test signal can be determined using the transfer characteristic determined in step S42. Further, the current pulsation component derived from the angle error is extracted by subtracting the amplitude and the phase of the current pulsation derived from the first test signal from the frequency analysis result determined in step S2 in relation to the current pulsation generated during the test operation. [0132] Next, the estimated angle error value is determined from the current pulsation component derived from the angle error, extracted in step S43, and the transfer characteristic at the frequency of the angle error, determined in step S42, using a similar method to the method described in the first embodiment (step S44) . [0133] In other words, values corresponding to the amplitude An and the phase §±1 of the first embodiment are determined from the current pulsation component derived from the angle error, extracted in step S43. Further, the amplitude and the phase of the angle error are determined from Equations (6) and (7) using the amplitude and phase determined in step S42 corresponding to AlL and ;t of the first embodiment. [0134] In the fourth embodiment of this invention, as described above, the motor 2 is operated in a condition where the frequencies of the test signals are aligned with the angle error frequency, and the angle error is estimated from the corresponding current pulsation component. At this time, the transfer characteristics determined during the operations performed by applying the test signals can be aligned precisely with the angle error frequency, and therefore the angle error can be estimated with an even higher degree of precision. Moreover, in contrast to the second embodiment, the motor speed remains the same during the two operations performed by applying the test signals, and therefore the cause of an estimation error in the angle error due to a difference in the operating speed can be eliminated. [0135] Further, the angle error may be estimated a plurality of times using the procedures described above with respect to a plurality of test signals having different initial phases and amplitude values, and an average value thereof may be set as the estimated angle error value. Moreover, this embodiment can be expanded easily to a case in which the angle error includes a plurality of frequency components. [0136] Note that instead of extracting the current pulsation component derived from the angle error in step S43, the motor 2 may be operated in a condition where the test signals are not applied, and the current pulsation component derived from the angle error may be determined by performing frequency analysis on the motor current at this time. [0137] Fifth Embodiment Fig. 12 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a fifth embodiment of this invention. In Fig. 12, elements marked with identical reference symbols Lo Fiy. 1 denote identical operations Lo the operations described in the first embodiment. [0138] In the fifth embodiment of this invention, an angle error correction device 5A is provided in place of the angle error correction device 5 shown in Fig. 1. The angle error correction device 5A includes the frequency analysis unit 51, an angle error estimation unit 52A, the angle error correction unit 53, and a resonance determination unit 54A. In other words, the angle error estimation unit 52A performs different operations to the angle error estimation unit 52 shown in Fig. 1, and the resonance determination unit 54A is provided additionally. [0139] The resonance determination unit 54A determines, on the basis of the frequency analysis result obtained by the frequency analysis unit 51 or the estimated angle error value obtained by the angle error estimation unit 52A, whether or not the angle error frequency of the position detector 3 or the frequency of the test signal matches a resonance frequency of the motor control system, and outputs a determination result to the angle error estimation unit 52A. [0140] When the motor 2 is connected to a load, the motor control, system may, depending on the dynamic characteristic of the load, have a resonance point. When the frequency of the angle error or the frequency of the test signal is close to or matches the frequency of the resonance point (i.e. the resonance frequency) during an operation of the motor 2, the precision with which the angle error is estimated may deteriorate. [0141] Hence, to avoid such a case, the angle error correction device 5A according to the fifth embodiment of this invention is capable of estimating the angle error with stability and a high degree of precision, as will be described below. An operation of the resonance determination unit 54A will be described below. The resonance determination unit 54A operates the motor 2 in order to determine whether or not the angle error frequency or the test signal frequency matches the resonance point before the angle error is estimated in the manner described in the first to fourth embodiments. [0142] Here, in a case where the resonance point does not vary according to the rotation position of the motor 2, for example when the load is a rotary device or the like, the motor 2 is operated while varying the operating speed thereof, and frequency analysis is performed on the motor current. [0143] Further, by performing frequency analysis on the motor current, a determination can be made as to whether or not the frequency of the angle error is in the vicinity of the resonance frequency on the basis of the values obtained in Equations (2) and (3) or Equations (4) and (5) . In the vicinity of the resonance frequency, the amplitude of the current pulsation increases or decreases rapidly, while the phase varies rapidly by close to 180 degrees. [014 4] Hence, a determination is made as to whether or not amounts of variation in the amplitude and phase of the current pulsation, determined by frequency analysis, exceed predetermined values, and when the amounts of variation exceed the predetermined values, the operating speed of the motor 2 on the periphery thereof is determined to be close to the resonance frequency. [014 5] Further, on the basis of the determination result obtained by the resonance determination unit 54A, the angle error estimation unit 52A outputs an operation command to the motor control device 1 to operate the motor 2 under conditions for avoiding the resonance frequency. [014 6] Furthermore, the angle error estimation unit 52A estimates the angle error using the methods described in the first to fourth embodiments. In this case, the operating speed of the motor 2 is modified so as to avoid the resonance frequency. When the operating speed of the motor 2 is modified, the frequency of the angle error and the frequency of the test signal change, and therefore the resonance frequency can be avoided. [0147] Hence, in the fifth embodiment of this invention, the resonance determination unit 54A determines whether or not the angle error frequency or the test signal frequency matches the resonance frequency of the motor control system, and estimates the angle error under conditions in which the angle error frequency and test signal frequency do not match the resonance frequency. As a result, the angle error can be estimated with stability and a high degree of precision. In particular, the resonance frequency can be avoided even when a load is attached, and therefore adjustments can be made with a high degree of precision during installation of the motor control system. [0148] In Ihe fifLh embodiment, an example in which Lhe angle error estimation unit 52A outputs an operation command to operate the motor 2 so as to avoid the resonance frequency was described, but instead, a control unit that advances the sequence of angle error estimation operations, including outputting the operation command for operating the motor 2, may be provided separately to the angle error correction device 5 and the motor control device 1 or as a dedicated control device. [0149] Sixth Embodiment Fig. 13 is a view showing a configuration of an elevator control device according to a sixth embodiment of this invention. Here, Fig. 13 is a view showing a case in which the motor control system including the angle error correction device for a position detector according to the first to fifth embodiments of this invention is applied to an elevator. In Fig. 13, parts marked with identical reference numerals to Fig. 1 or Fig. 12 denote identical operations to the operations described in the first to fifth embodiments. [0150] In Fig. 13, a car 7 and a counter weight 9 of an elevator are connected to each other by a hoisting rope 8 and suspended from a sheave 6 in the manner of a well bucket. The sheave 6 is connected to the motor 2, which serves as a motor for driving the car 7, and the car 7 is raised and lowered by power from the motor 2. [0151] Here, the angle error is estimated during installation of a hoisting machine, for example. More specifically, the angle error is estimated while rotating the hoisting machine by installing the molur 2 in an elevaLox system as Lhe hoisting machine and performing an operation for estimating the angle error in a condition where the rope 8 either is or is not wound around the sheave 6. [0152] At this time, the angle error can be estimated with stability by estimating the angle error only during an interval in which the car 7 travels at a constant speed. Further, to extend the interval in which the car 7 travels at a constant speed, an operation may be performed after setting a travel speed to be lower than a rated speed of the elevator. Moreover, to improve the estimation precision, the travel speed of the elevator may be modified to a travel speed at which the amplitude of the current pulsation increases. Note that there are no limitations on the position of the car 7, and the angle error may be estimated in any position within a hoistway through which the car 7 travels. [0153] Further, to increase the amplitude of the current pulsation in order to improve the estimation precision, an operation may be performed after modifying respective gains of the speed control unit and the position control unit so that the gains increase. In the case of PID control, a proportional gain, an integral gain, and a differential gain correspond to the gains of the control device. [0154] Furthermore, the angle error estimation result is recorded in a storage medium (a non-volatile memory, for example) as an angle error corresponding to a magnetic pole position of the hoisting machine. During a normal operation, the estimated angle error value COrrespundiny Lo the uuLpuL of the posiLion delecloi 3 is read from the storage medium and corrected. Information relating to the angle error recorded in the storage medium may be information that indicates the error amplitude and the phase shift of the angle error and is used to determine the angle error by solving Equation (1), or may be corrected angle information or corrected position information that corresponds to the magnetic pole position of the hoisting machine and is provided in the form of a table or the like. In this case, a method of storing phase information and amplitude information in advance and correcting the angle error by calculation is preferably employed in order to minimize the amount of information. [0155] Note that in the elevator, the dynamic characteristic of the elevator system varies according to the position and the load weight of the car 7, and therefore the transfer characteristic shown in Fig. 6 likewise varies according to the position and load weight of the car 7. Accordingly, when the angle error is estimated by performing operations in which correction signals are applied a plurality of times, the operations are preferably performed under conditions in which the position and load weight of the car remain constant or nearly constant. [0156] Further, in the elevator, the dynamic characteristic of the elevator system varies when specifications such as an ascent/descent length and a rated load capacity vary, but in this invention, the transfer characteristic of the motor control system is determined during an operation performed by applying a test signal, and therefore the angle error can be estimated regardless of the specifications of the elevator. Needless to mention, this invention is not limited to an elevator, and may be used to estimate an angle error in any system in which a load characteristic of a motor varies from moment to moment. [0157] Furthermore, in this invention, the angle error can be estimated simply by performing frequency analysis in a minimum of two patterns, and therefore the angle error can be estimated quickly. Moreover, once estimation has begun, estimation is performed continuously without stopping the motor 2, and therefore the angle error can be estimated quickly. Hence, the angle error can be estimated quickly during a test operation following installation of the elevator, for example, meaning that there is no need to secure time for estimating the angle error. As a result, the amount of time required for adjustments during installation can be reduced. [0158] Next, a case in which the angle error is estimated while varying the position of the car 7 will be described. In a case where the angle error is estimated while operating the car 7 from a lowermost floor to an uppermost floor or from the uppermost floor to the lowermost floor during installation, for example, the angle error can be estimated with a high degree of precision by implementing following procedures. [0159] In the elevator, resonance points derived from an elastic characteristic of the rope 8 exist between the car 7 and the rope 8 and between the counter weight 9 and the rope 8 . Further, the resonance points vary according to the position of the car 7 and the load weight of the car. Therefore, the period of the periodic angle error of the position detector 3 and the frequency of the test signal used to estimate the angle error may match the resonance frequencies of the resonance points. Here, when the frequency of the angle error or the frequency of the test signal matches a resonance frequency of the elevator, the amplitude and phase of the current value used during frequency analysis vary rapidly, leading to instability in the frequency analysis result, and as a result, the precision with which the angle error is estimated deteriorates. [0160] Hence, before estimating the angle error, the car 7 of the elevator is operated from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor, and frequency analysis is performed on the motor current at the frequency corresponding to the angle error. When, at this time, the frequency of the angle error is in the vicinity of the resonance frequency, the amplitude of the corresponding current pulsation increases or decreases rapidly, while the phase thereof varies rapidly by close to 180 degrees. [0161] Hence, a determination is made as to whether or not the amounts of variation in the amplitude and phase of the current pulsation, determined by frequency analysis, exceed predetermined values, and when the amounts of variation exceed the predetermined values, the frequency of the angle error on the periphery thereof is determined to be close to the resonance frequency. Further, on the basis of the determination result, the angle error is estimated in a differenl position Lo the position determined to be close to the resonance frequency. Note that the operating speed during angle error estimation may be modified so as not to approach the resonance frequency. Further, the method described above is not limited to an elevator, and may also be applied to a case in which the resonance frequency varies according to the rotation position of the motor 2. [0162] For example, the angle error may be estimated during installation of the elevator as follows. First, the car 7 of the elevator is operated from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor, frequency analysis is performed on the motor current at the frequency corresponding to the angle error, and the amounts of variation in the amplitude and phase of the current pulsation are calculated. [0163] At this time, the car position is stored together with the amounts of variation in the amplitude and phase of the current pulsation. Next, when the operation from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor is complete, a determination is made as to whether or not the amounts of variation in the amplitude and phase of the current-pulsation exceed the predetermined values, and positions in which the amounts of variation do not exceed the predetermined values are extracted. Next, the car 7 is moved to a position in which the amounts of variation in the amplitude and phase of the current pulsation do not exceed the predetermined values, whereupon angle error estimation is implemented. [0164] When the upeidLiun performed to determine wheLher or not the amounts of variation in the amplitude and phase of the current pulsation exceed the predetermined values is an operation from the lowermost floor to the uppermost floor and the operation for estimating the angle error is implemented in the opposite direction, i.e. from the uppermost floor to the lowermost floor, the angle error can be estimated during a single reciprocating operation, and as a result, the amount of time relating to the angle error estimation can be reduced. [0165] On the other hand, when the operation performed to determine whether or not the amounts of variation in the amplitude and phase of the current pulsation exceed the predetermined values is an operation from the uppermost floor to the lowermost floor, the operation for estimating the angle error may be implemented in the opposite direction, i.e. from the lowermost floor to the uppermost floor. With this estimation method, the angle error can be estimated while avoiding deterioration of the estimation precision due to resonance, and therefore the angle error can be corrected accurately. Moreover, the angle error can be estimated accurately during a single reciprocating operation, and as a result, the amount of time required for adjustments during installation is reduced. [0166] Note that the overall device layout, the roping system, and so on of the elevator are not limited to the example shown ir Fig. 13. For example, this invention may be applied to an elevator employing a 2:1 roping system. Further, the position of the hoisting machine constituted by Lhe inoLui 2, fur example, is nuI limited to the example shown in Fig. 13. Moreover, this inventior may be applied to various types of elevators, such as a machine room-less elevator, a double deck elevator, a one-shaft, multi-cai elevator, and an inclined elevator, for example. DESCRIPTION Title of Invention: ANGLE ERROR CORRECTION DEVICE AND ANGLE ERROR CORRECTION METHOD FOR POSITION DETECTOR, ELEVATOR CONTROL DEVICE, AND ELEVATOR SYSTEM Technical Field [0 001] This invention relates to an angle error correction device and an angle error correction method for a position detector, which are applied to a control device for an elevator hoisting machine, a control device for an in~vehicle motor, a control device for a motor of a machine tool, or the like, for example, in order to correct an angle error in a position detector having a periodic error that is determined univocally in accordance with a rotation position of a motor, and also to an elevator control device and an elevator system. Background Art [0002] In a conventional resolver angle detection device, an angle detector detects an angle signal from a signal detected in a resolver, whereupon an angle error estimator calculates and corrects an angle error using the fact that an error waveform of the resolver is constituted by an n-th order component that is unique to the resolver and reproducible (see PTL 1, for example). [0003] In the resolver angle detection device of PTL 1, a position error is calculated by referring to the detected angle signal, a speed error signal is calculated by differentiating the 1 position error, and a detected error is calculated for each frequency component by performing frequency analysis on the speed error signal using the Fourier transform, for example. Further, an estimated angle error signal is generated by combining the calculated angle errors, and the detected angle signal is corrected by an angle signal correction circuit using the generated estimated angle error signal. Citation List Patent Literature [0004] [PTL 1] Japanese Patent Application Publication No. 2012-145371 Summary of Invention Technical Problem [0005] However, the prior art includes the following problem. In the conventional resolver angle detection device, a motor rotation speed is detected by a speed detector from the angle signal detected by the angle detector, and the angle error is estimated using the detected speed. Here, when the angle error is estimated using the detected speed, the estimation precision of the angle error is determined according to a speed resolution of the angle detector or the speed detector. When the angle detector or the speed detector has a low speed resolution, therefore, a quantization error occurs, and as a result, the angle error cannot be estimated with a sufficient degree of precision. [0006] This invention has been designed to solve the problem described above, and an object thereof is to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately. Solution to Problem [0007] An angle error correction device for a position detector according to this invention is used in combination with a motor control device that controls a motor, a position detector that outputs a position detection signal obtained by detecting a rotation position of the motor, and a current detector that detects a current flowing through the motor to correct a periodic angle error that is included in the position detection signal and determined in accordance with the rotation position, the angle error correction device including: a frequency analysis unit that performs frequency analysis on a detected current detected by the current detector while rotating the motor, calculates an amplitude and a phase of a specific frequency, and outputs the amplitude and the phase of the specific frequency as a frequency analysis result; an angle error correction unit that outputs an addition signal, the addition signal being obtained by adding the position detection signal to an input signal input therein, to the motor control device; and an angle error estimation unit that implements first control processing and second control processing repeatedly on a plurality of different test signals, the first control processing being processing for inputting a set value of a test signal having known amplitude, phase, and frequency values into the angle error correction unit as the input signal and having the angle error correction unit operate the motor by applying a test signal corresponding to the set value thereto, the second control processing being processing lor having the frequency analysis unil perform frequency analysis at the frequency of the test signal on the detected current obtained in the first control processing, estimates estimated values of an amplitude and a phase of the angle error on the basis of amplitudes and phases constituting at least two types of frequency analysis results calculated by the frequency analysis unit in the second control processing, and outputs the estimated values to the error correction unit, wherein the angle error correction unit outputs the addition signal to the motor control device using the estimated values of the amplitude and the phase of the angle error as the input signal. [0008] Further, an angle error correction method for a position detector according to this invention is executed by an angle error correction device for a position detector, the angle error correction device being used in combination with a motor control device that controls a motor, a position detector that outputs a position detection signal obtained by detecting a rotation position of the motor, and a current detector that detects a current flowing through the motor to correct a periodic angle error that is included in the position detection signal and determined in accordance with the rotation position, the angle error correction method including: a frequency analysis step in which frequency analysis is performed on a detected current detected by the current detector while rotating the motor, an amplitude and a phase of a specific frequency are calculated, and the amplitude and the phase of the specific frequency are output as a frequency analysis resulL; an angle error correction step in which an addition signal obtained by adding the position detection signal to an input signal input therein is output to the motor control device; a first control step in which a set value of a test signal having known test amplitude, phase, and frequency values is input as the input signal during the angle error correction step, and the motor is operated by applying a test signal corresponding to the set value thereto; a second control step in which frequency analysis is performed during the frequency analysis step at the frequency of the test signal on the detected current obtained in the first control step; the first control step and the second control step being executed repeatedly on a plurality of different test signals, a third control step in which estimated values of an amplitude and a phase of the angle error are estimated on the basis of amplitudes and phases constituting at least two types of frequency analysis results calculated in the frequency analysis step during the second control step; and a fourth control step in which the addition signal is output to the motor control device using the estimated values of the amplitude and the phase of the angle error as the input signal during the angle error correction step. [0009] Further, an elevator control device according to this invention includes a motor control device that controls a hoisting machine of an elevator, a position detector that detects a rotation position of the hoisting machine, and includes a periodic error determined univocally in accordance with the rotation position, a current detector that detects a current flowing through the hoisting machine, and an angle error correction device that is connected to the motor control device, the position detector, and the current detector. Advantageous Effects of Invention [0010] With the angle error correction device and angle error correction method for a position detector according to this invention, the position detector can perform frequency analysis on a specific frequency component of the motor current, and estimate the periodic error of the position detector on the basis of the frequency analysis result. At this time, processing for performing an operation by applying the test signal having known amplitude, phase, and frequency values and performing frequency analysis at the frequency of the test signal is implemented a plurality of times, whereupon the error of the position detector is estimated on the basis of amplitudes and phases calculated in the plurality of frequency analyses. As a result, it is possible to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately. [0011] Further, with the elevator control device according to this invention, the position detector can perform frequency analysis on a specific frequency component of the motor current, and estimate the periodic error of the position detector on the basis of the frequency analysis result. At this time, processing for performing an operation by applying a test signal having known amplitude, phase, and frequency values and performing frequency analysis at the frequency of the test signal is implemented a plurality of times, whereupon the error of the position detector is estimated on the basis of amplitudes and phases calculated in the plurality of frequency analyses. As a result, it is possible to obtain an angle error correction device and an angle error correction method for a position detector, with which an angle error can be estimated and corrected accurately in an elevator system. Brief Description of Drawings [0012] Fig. lis a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a first embodiment of this invention. Fig. 2 is a block diagram showing a configuration of a motor control device of the motor control system shown in Fig. 1. P^ig. 3 is an illustrative view showing a relationship between a motor rotation angle and a detected angle including a periodic error in a position detector to which the angle error correction n device for a position detector according to the first embodiment of this invention is applied. Fig. 4 is a block diagram illustrating an output of an angle error correction unit in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 5 is a block diagram illustrating a transfer characteristic from the output of the angle error correction unit to a motor current in the form of a transfer function expression in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 6 is a Bode diagram showing an example of the transfer characteristic from the output of the angle error correction unit to the motor current in the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 7 is a flowchart showing processing executed by an angle error estimation unit of the angle error correction device for a position detector according to the first embodiment of this invention. Fig. 8 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a second embodiment of this invention. Fig. 9 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a third embodiment of this invention. Fig. 10 is a flowchart showing processing executed by an angle error estimation unit of an angle error correction device for a position detector according to a fourth embodiment of this invention. Fig. 11 is a flowchart showing in detail processing for calculating an estimated angle error value, which is executed by the angle error estimation unit of the angle error correction device for a position detector according to the fourth embodiment of this invention. Fig. 12 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a fifth embodiment of this invention. Fig. 13 is a view showing a configuration of an elevator control device according to a sixth embodiment of this invention. Description of Embodiments [0013] Preferred embodiments of an angle error correction device and an angle error correction method for a position detector according to this invention will be described below using the drawings. Note that identical or corresponding parts of the drawings will be described using identical reference numerals. [0014] First Embodiment Fig. 1 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a first embodiment of this invention. In Fig. 1, the motor control system includes a motor control device 1, a motor 2, a position detector 3, a current detector 4, and an angle error correction device 5. [0010] The motor control device 1 is a device for controlling a rotation speed or a rotation position of the motor 2. Here, referring to Fig. 2, a case in which the motor control device 1 controls the rotation speed of the motor 2 will be described. Fig. 2 is a block diagram showing a configuration of the motor control device provided in the motor control system shown in Fig. 1. [0016] In Fig. 2, the motor control device 1 includes a speed calculation unit 11, a speed control unit 12, a current control unit 13, and a power converter 14. The speed calculation unit 11 calculates the rotation speed of the motor 2 on the basis of position information (a corrected rotation position of the motor 2) or angle information obtained when the angle error correction device 5 corrects the rotation position (a motor rotation position) of the motor 2, detected by the position detector 3. Note that the speed calculation unit 11 calculates the rotation speed most simply by time-differentiating the position information or the angle information. [0017] The speed control unit 12 calculates a current command value (a torque command value) for the motor 2 so that the rotation speed of the motor 2 reaches a desired speed. The current control unit 13 calculates a voltage command value for the motor 2 so that a current (a motor current) of the motor 2, detected by the current detector 4, matches the current command value output by the speed control unit 12. The power converter 14 applies an applied motor voltage to the motor 2 on the basis of the voltage command value output by the current control unit 13 in order to control the motor current. [0018] Here, P control, PI control, and PID control are typically used as the control executed by the speed control unit 12 and the current control unit 13, but various other control methods may be used. Further, vector control is typically used to control the current of the motor 2, in which case the motor current and the applied motor voltage are converted to d-q axes, whereupon a control method such as the aforesaid PID control is applied to the converted current and voltage. [0019] Furthermore, an inverter is typically used as the power converter 14 that applies a voltage to the motor 2. The inverter converts a voltage from a power supply, not shown in the drawings, into a desired variable-voltage, variable-frequency voltage. In this invention, a power converter that converts an AC voltage into a DC voltage using a converter and then converts the DC voltage into an AC voltage using an inverter, such as a typical, commercially available inverter device, or a variable-voltage, variable-frequency power converter including a power converter that converts an AC voltage directly into a variable-voltage, variable-frequency AC voltage, such as a matrix converter, may be used. [0020] Further, the inverter according to the first embodiment of this invention may include a coordinate conversion function in addition to the inverter function described above. More specifically, when the voltage command value is a voltage command value on d-q axes, the inverter is assumed to include a coordinate conversion function for converLing the vollage command vdlue on the d-q axes into a voltage corresponding to the instructed voltage command value by converting the voltage command value on the d-q axes into a phase voltage or a line voltage. Note that this invention may be applied likewise when a device or means for correcting a dead time of the inverter is provided. [0021] Moreover, when the rotation position of the motor 2 is controlled, a position control unit is added on a higher order than the speed control unit 12. The position control unit calculates a speed command value for the motor 2, which is input into the speed control unit 12 so that the rotation position of the motor 2 reaches a desired position. The speed control unit 12 executes the control described above using the speed command value as a desired speed. P control, PI control, and PID control are typically used as the control executed by the position control unit, but various other control methods may be used. [0022] Alternatively, the speed control unit 12 may be omitted such that the motor control device is constituted by the position control unit and the current control unit 13. In this case, the position control unit calculates the current command value for the motor 2 so that the rotation position of the motor 2 reaches the -i o desired position. [0023] Returning to Fig. 1, the current detector 4 measures a current of the motor 2. When the motor 2 is a three-phase motor, for example, a two-phase phase current is typically measured. However, a three-phase phase current may be measured instead. [0024] Further, the position detector 3, which is constituted by an optical encoder, a magnetic encoder, or a resolver, for example, detects the rotation position of the motor 2, which is required to control the motor 2, and outputs a position detection signal. Furthermore, as shown in Fig. 3, rotation position information output by the position detector 3 includes a periodic error determined univocally in accordance with the rotation position of the motor 2. [0025] Here, a periodic error determined univocally in accordance with the rotation position of the motor 2 is an error that can be reproduced in accordance with the rotation position, such as the detection error of the resolver described in paragraphs 0020 and 0021 of PTL 1, or a missing pulse or an imbalance in an inter-pulse distance caused by a slit defect in an optical encoder, for example. [0026] Hereafter, the periodic error determined univocally in accordance with the rotation position of the motor 2 will be expressed as an angle error 0err that is obtained by converting the position information into an angle. Note that this invention may be applied to a case in which the position detector 3 includes a periodic error determined univocally in accordance with the rotation position of the motor 2, and a main component order of the angle error Gerr is known. [0027] The periodic angle error 0crr of the position detector 3 can be expressed approximately using a sine wave, as shown below in Equation (1) . Note that since there is essentially no difference between sine wave notation and cosine wave notation, sine wave notation is used throughout the first embodiment of this invention. [0028] [Math.1] 0err * 4 sm(NxOm + (Px) + A2 sm(N2&m + „)... ( 1) [0029] In Equation (1), 0m denotes a mechanical angle of the motor 2, A! denotes an error amplitude of an Nx-th order, A2 denotes an error amplitude of an N2-th order, An denotes an error amplitude of an Nn~th order, §i denotes a phase shift (an initial phase) of the Ni-th order relative to the mechanical angle of the motor 2, 2 denotes a phase shift of the N2-th order relative to the mechanical angle of the motor 2, and (|>n denotes a phase shift of the Nn-th order relative to the mechanical angle of the motor 2. [0030] Note that the spatial orders Ni, N2, . . . , Nn in Equation (1) are spatial orders of a main component of the periodic error determined univocally in accordance with the rotation position of the motor 2, and need not be consecutive integers such as 1, 2, . . . , Nn. Here, the main component is a component in which an amplitude of the component at the corresponding spatial order is larger than amplitudes at other frequencies. 14 [0031] Further, in Equation (1) , three or more frequency components are combined, but the frequency components of the periodic angle error 0err may be constituted by one, two, or more components . [0032] In the first embodiment of this invention, the angle error correction device 5 includes a frequency analysis unit 51, an angle error estimation unit 52, and an angle error correction unit 53. The angle error estimation unit 52 estimates the periodic angle error of the position detector 3 as a function of Equation (1) on the basis of a frequency analysis result obtained when the frequency analysis unit 51 analyzes the motor current. [0033] Further, the angle error correction unit 53 generates an addition signal by adding the position detection signal from the position detector 3 to an input signal from the angle error estimation unit 52. More specifically, the angle error correction unit 53 generates an angle error correction value in an identical format to Equation (1) on the basis of an angle error estimation result, and corrects the angle error by adding the generated angle error correction value to the information indicating the rotation position or the angle of the motor 2. [003 4] An operation of the angle error correction device 5 will now be described. The frequency analysis unit 51 calculates at least one of the amplitude and the phase of the motor current at a specific frequency on the basis of the motor current detected by the current detector 4 and the information indicating the rotation position or the angle of the motor 2, i.e. the output of the position detector 3. [0035] Note that although the information indicating the rotation position or the angle of the motor 2 is input into the frequency analysis unit 51 in Fig. 1, the invention is not limited to this configuration, and position information or angle information obtained when the angle error correction unit 53 corrects the rotation position of the motor 2 may be input instead. [0036] Here, the frequency analysis unit 51 is preferably configured to obtain the amplitude and phase at a desired frequency of the input signal by means of Fourier transform, Fourier series analysis, or fast Fourier transform, but may be configured to extract a desired frequency signal by means of a filter combining a notch filter and a band pass filter, and then calculate the desired amplitude and phase of the input signal using an amplitude detection unit and a phase detection unit. Further, the filter used in this case may be an electric filter combining a resistor, a capacitor, a coil, and so on, or may take the form of processing executed by a computer. [0037] Furthermore, the motor current input into the frequency analysis unit 51 may be any current used in vector control, such as a d axis current, a q axis current, a y axis current, a 5 axis current, an a axis current, or a (3 axis current, these currents being obtained by subjecting a phase current to coordinate conversion. [0038] Note that here, a signal having a desired frequency (a specific frequency) denotes a signal that is derived from the periodic angle error 0err of the position detector 3 and has an identical frequency to the main component of the angle error 9err/ or a signal having an identical frequency to the main component of a test signal generated by the angle error correction unit 53. Note that the test signal will be described in detail below. Further, in the first embodiment of this invention, the desired frequency is expressed as a spatial frequency, but a temporal frequency is essentially no different thereto. [0039] Here, the spatial frequency denotes a frequency within a specific interval, which in the first embodiment of this invention corresponds to a single rotation of the motor 2 . Moreover, a signal including N periodic waves within a single mechanical angle rotation of the motor 2 will be referred to as a wave of the spatial order N. [0040] In the control device for the motor 2 having the position detector 3, the error of the position detector 3 occurs periodically in accordance with the rotation position of the motor 2, and therefore frequency analysis preferably involves analysis by means of spatial frequency. Likewise in Equation (1) , the angle error Gerr is expressed in terms of the spatial frequency, and moreover, the inputs of the frequency analysis unit 51 shown in Fig. 1 are inputs (a current and an angle) corresponding to spatial frequency analysis. [0041] However, the first embodiment of this invention can also be applied to frequency analysis by means of temporal frequency, and when frequency analysis is performed by means of temporal frequency, the frequency analysis is performed using the detected angle, a measured time measured by a time measurement unit, and the current as inputs in place of the current and the angle. [0042] The angle error estimation unit 52 estimates the angle error 9err determined univocally in accordance with the rotation position of the motor 2 by means of an estimation method to be described below using a current amplitude value, or a current amplitude value and a phase value, of the desired frequency component, these values serving as the output of the frequency analysis unit 51, and either the rotation position of the motor 2, which serves as the output of the position detector 3, or information indicating the angle corrected by the angle error correction unit 53 . The angle error estimation unit 52 then outputs the estimated angle error value to the angle error correction unit 53 as angle information or position information. [0043] The angle error correction unit 53 adds an angle error correction signal based on the estimated angle error value serving as the output of the angle error estimation unit 52 to the rotation position of the motor 2, which serves as the output of the position detector 3, and outputs corrected position information or angle information. [0044] A case in which the estimated angle error value is output in an identical unit system to the output signal of the position detector 3 will now be described as an example. When the position detector 3 is an optical encoder having a resolution of 1024 pulses per rotation, and the estimation result obtained by the angle error estimation unit 52 is 1°, the angle error estimation unit 52 outputs three pulses, which is the number of pulses corresponding to 1°, as the position information. [0045] As shown in Equation (1) , when the angle error includes a plurality of frequency components, the angle errors of the respective components may be estimated in succession and then added together, or the plurality of frequency components may be estimated simultaneously. By employing simultaneous estimation at this time, an estimation time can be shortened in comparison with a case where the angle errors of the respective components are estimated in succession. Here, for simplicity, a case in which the angle error includes only one frequency component will be described. [0046] It is known that when speed feedback control or position feedback control is implemented by the position detector 3 that includes a periodic angle error determined univocally in accordance with the rotation position of the motor 2, current pulsation or current command value (torque command value) pulsation that includes a frequency component of the same order as the angle error is generated. [0047] Similarly, it is known that when speed feedback control or position feedback control is implemented by having the angle error correction unit 53 generate a specific periodic signal and adding the generated specific period signal to the output of the position detector 3, current pulsation or current command value (torque command value) pulsation that includes an identical frequency component to the added signal is generated. [0048] When, dL Lhis Lime, the frequency analysis unit 51 performs frequency analysis on the phase current in a case where the motor 2 is a permanent magnet synchronous motor, the spatial order of the current pulsation appearing in the phase current is a Pn ± Nn-th order, where Pn denotes a number of pole pairs and Nn denotes the order of the desired frequency. [0049] It is therefore sufficient to perform frequency analysis on the current of at least one phase of the phase current, and estimate a Pn + Nn-th order or Pn - Nn-th order angle error from a Pn + Nn-th order or Pn - Nn-th order current. Note, however, that when the order Nn of the desired frequency is larger than the number of pole pairs Pn of the motor 2, the Pn - Nn-th order may take a negative value so as to become non-existent, and therefore, in this case, frequency analysis is performed on the Pn + Nn~th order current. Moreover, a constant-torque, constant-speed operation is preferably implemented when estimating the angle error. [0050] Further, when the frequency analysis unit 51 performs frequency analysis on either a d axis current or a q axis current and the order of the desired frequency is Nn, current pulsation components appearing on the dq axes have pulsation components of an identical order to the Nn-th order. Furthermore, the q axis current, which is a torque current, flows around the d axis current due to a magnetic pole deviation caused by position pulsation of the desired frequency, and therefore the d axis current exhibits current pulsation analogous to the angle error. Moreover, speed pulsation in Ihe q axis currenL forms current coruruand value (a torque command value) pulsation via a speed control system. Accordingly, the q axis current forms current pulsation analogous to the angle error that causes the speed pulsation. [0051] Note that when frequency analysis is performed using a current detection value or a current command value fa torque command value) of the d axis current or the q axis current, estimation is performed in a condition where the q axis current to be fed back is fixed, or in other words in a condition of constant acceleration. In particular, estimation is preferably performed in a condition where the acceleration is zero, or in other words a condition in which the motor 2 rotates at a constant speed. [0052] The angle error and the signal from the angle error correction unit 53 generate current pulsation in accordance with a transfer characteristic determined according to dynamic characteristics of the motor control device 1, the motor 2, and a load connected to the motor 2. Hence, when the transfer characteristic can be determined, it is possible to estimate the angle error signal by which the current pulsation is generated. In other words, the angle error by which the current pulsation is generated can be determined by back calculation from the determined transfer characteristic and the current pulsation. [0053] A method of estimating the transfer characteristic and the periodic angle error component from a frequency analysis result obtained in relation to a current pulsation component will now be described. Note that frequency analysis may be performed on a current command value (torque command value} pulsation component instead of the current pulsation component, but a case in which frequency analysis is performed on the current pulsation component will be described below. [0054] Fig. 4 is a block diagram illustrating the output of the angle error correction unit in the angle error correction device for a position detector according to the first embodiment of this invention. In Fig. 4, the angle error correction unit 53 generates an angle error correction signal. The corrected rotation position of the motor 2, which is corrected by adding the angle error correction signal generated by the angle error correction unit 53 to the motor rotation position serving as the output of the position detector 3, is fed back to the motor control device 1. The motor current detected by the current detector 4 is also fed back to the motor control device 1. [0055] When, at this time, a transfer characteristic from the angle error correction signal to the motor current is represented in the form of a transfer function expression as Gerr i (s) , a block diagram thereof is as shown in Fig. 5. Fig. 5 is a block diagram illustrating the transfer characteristic from the output of the angle error correction unit to the motor current in the form of a transfer function expression in the angle error correction device for a position detector according to the first embodiment of this invention. [0056] Here, "s" is a Laplace operator. Further, Gerr_i (s) also matches a transfer function from the motor rotation position, which serves as the output of the position detector 3, to the motor current. Note that when a load is connected to the motor 2, the dynamic characteristic of the load is also represented by Gerr_i (s) . [0057] In this invention, a gain and a phase of Gerr_i (s) at the frequency of the angle error or a specific frequency are determined, whereupon the angle error is estimated from the determined gain and phase. Fig. 6 shows an example of Gerr_i (s). Fig. 6 is a Bode diagram showing an example of the transfer characteristic from the output of the angle error correction unit to the motor current in the angle error correction device for a position detector according to the first embodiment of this invention. [0058] In Fig. 6, an upper section shows a gain characteristic and a lower section shows a phase characteristic. When the angle error frequency varies, the amplitude and the phase of motor current pulsation corresponding to the angle error vary in accordance with the characteristics shown in Fig. 6. Further, the angle error frequency varies according to the rotation speed of the motor 2. In other words, the phase and the amplitude of current pulsation derived from the angle error vary in accordance with the rotation speed. [0059] Next, processing executed by the angle error estimation unit 52 according to the first embodiment of this invention will be described with reference to a flowchart shown in Fig. 7. [0060] First, at the start of angle error estimation, the angle error estimation unit 52 outputs an operation command for operating Lhe motor 2 Lo Lhe motor control device 1, cuid outputs a command to set the angle error correction signal at zero, or in other words a command to set the test signal at zero, to the angle error correction unit 53 (step SI). As a result, the angle error correction unit 53 sets the angle error correction signal at zero such that the motor 2 is rotated in a condition where angle error correction is not performed. [00 61] Next, the angle error estimation unit 52 outputs a frequency analysis command to the frequency analysis unit 51 to cause the frequency analysis unit 51 to perform frequency analysis on the motor current (step S2). The frequency analysis result is input into the angle error estimation unit 52. At this time, frequency analysis is performed on the motor current at a frequency corresponding to the angle error frequency. [0062] This is achieved by determining Fourier coefficients of the current pulsation corresponding to the specific frequency, for example. A case in which the Fourier coefficients of the current pulsation are determined with respect to the q axis current will be described below. The Fourier coefficients at a frequency Mi [Hz] of the current pulsation of the q axis current, which corresponds to the angle error frequency, can be determined by solving arithmetic expressions in Equations (2) and (3), shown below. [0063] [Math.2] r 2 n , , Ki = ™\ iq(t)cos(2nM1t) dt • • • (2) T J_z 2 [0064j [Math.JJ r 2 fz , , Sm=^ ig(t)sin(27rM1t) dt • * * (3) [0065] In Equations (2) and (3), iq (t) denotes the value of the q axis current, and T denotes a current pulsation period of the frequency Mi [Hz] . Note that T = 1 /Mi. Further, Anl and Bni denote a cosine wave coefficient and a sine wave coefficient, respectively. [0066] Equations (2) and (3) show a case in which the Fourier coefficients are determined by being integrated over time, but the Fourier coefficients may be determined by being integrated in accordance with the rotation angle of the motor 2. Further, Equations (2) and (3) are arithmetic expressions in a continuous time region, but when Equations (2) and (3) are packaged in a computer such as a microcomputer, the equations are packaged after being converted into expressions in discrete time regions. Furthermore, Equations (2) and (3) can be solved using cosine wave and sine wave signal generators, a multiplier, and an integrator, and can therefore be packaged in a computer easily. [0067] Moreover, in Equations (2) and (3), the Fourier coefficients are calculated by being integrated over a single signal period, but may be determined as a value that is integrated over a number of periods and then divided by the number of periods. In this case, the Fourier coefficients are determined as an average value of the number of periods, and therefore the effects of irregularities and disturbances on the current pulsation can be reduced. Furthermore, an inLeyraLion start Lime preferably slarls from a reference point (zero degrees, for example) of the rotation angle of the motor 2. As a result, the Fourier coefficients can be determined using the rotation angle of the motor 2 as a reference. [0068] Here, an amplitude An and a phase (j)n of the current pulsation component can be determined from the Fourier coefficients of Equations (2) and (3) using Equations (4) and (5), shown below. [0069] [Math.4] An = ^A2nl + Bl± • - ■ (4) [0070] [Math.5] n determined in Equations (4) and (5) . Note that instead, the Fourier coefficients Ani and Bni may be stored, and the amplitude and phase may be determined by solving Equations (4) and (5) . [0072] Next, the angle error estimation unit 52 operates the motor 2 by applying the test signal thereto (step S3) such that frequency analysis is performed on the motor current in a condition where the motor 2 is rotated (step S4) . The rotation speed of the motor 2 at this time is set at an identical speed to step SI. Note that an operation command for operating the motor 2 and a set value of the test signal are output by the angle error estimation unit 52. Further, the frequency analysis result is input into the angle error estimation unit 52. [0073] Here, the test signal is a sine wave or cosine wave test signal set at an amplitude, a frequency, and an initial phase that have been determined in advance. The test signal is generated by the angle error correction unit 53 and added to the output of the position detector 3. A sine wave and a cosine wave can be converted into each other by varying the respective initial phases thereof, and therefore a sine wave will be used in the following description. Furthermore, the predetermined amplitude of the test signal is set at At, and the initial phase thereof is set at (|>t. [0074] In the first embodiment of this invention, a sine wave signal having a different frequency to the angle error frequency is applied as the test signal. For example, a sine wave signal having a frequency in the vicinity of the angle error frequency is used as the test signal. Note that the vicinity denotes frequencies that are larger or smaller than the angle error frequency by approximately 10% to 20%, for example, so that the gain and the phase at the frequency of the test signal are within a range that may be considered substantially equal to the gain and the phase of the angle error frequency. [0075] By setting the frequency of the test signal at a different value to the angle error frequency in this manner, the test signal and the angle error correction signal do not intersect, and therefore the frequency analysis can be performed easily, thereby facilitating calculation of the transfer characteristic and estirnation of Ihe angly exioi. [0076] Further, when frequency analysis is performed on the motor current in step S4, the frequency analysis is performed at the current pulsation frequency corresponding to the test signal applied in step S3. Note that when the frequency analysis is performed on the d axis current or the q axis current, an identical frequency to the frequency of the test signal is obtained. The frequency analysis is performed using similar calculations to those shown in Equations (2) to (5) , but the values of Mi and T are replaced with a frequency and a period corresponding to the test signal. At this time, the amplitude and the phase of the current pulsation, determined from the Fourier coefficients of the current pulsation, are set respectively at Ait and §n. [0077] Next, the angle error estimation unit 52 calculates the estimated angle error value of the position detector 3 (step S5}. In step S4, the current pulsation amplitude corresponding to the amplitude At of the test signal is Alt, and it is therefore evident that the current pulsation is obtained by multiplying the amplitude of the test signal by hn/At. [0078] Here, when the angle error frequency and the frequency of the test signal are close, a magnification ratio of the current pulsation amplitude of the current pulsation generated by the angle error relative to the angle error amplitude may be considered identical to the aforesaid magnification ratio, and as a result, the amplitude of an error signal at which the current pulsation amplitude An is yenerctted can be determined in relation to the error signal determined in step S2. In other words, when the amplitude of the error signal is set at AL, the amplitude can be determined from Equation (6), shown below. [0079] [Math.6] Ai=^At ■ • • (6) [0080] Further, in step S4, the current pulsation phase corresponding to the initial phase t of the test signal is it, and it is therefore evident that the current pulsation phase deviates from the phase of the test signal by ii is generated can be determined in relation to the error signal determined in step S2 . In other words, when the phase of the error signal is set at <|>i, the phase can be determined from Equation (7), shown below. [0082] [Math.7] .it of the first embodiment are determined using the Fourier coefficient of the current pulsation component derived from the extracted test signal, whereupon the estimated angle error value is determined from Equations (6) and (7) using An and (j)j_i determined in step S2. Note that At and t in Equations (6) and (7) respectively denote the amplitude and the initial phase of the test signal, and similarly to the first embodiment are known. [0118] in the third embodiment of this invention, as described above, the motor 2 is operated in a condition where the frequency of the test signal matches the angle error frequency, and the angle error is estimated from the corresponding current pulsation component. At this time, the transfer characteristic determined during the operation performed by applying the test signal can be aligned precisely with the angle error frequency, and therefore the angle error can be estimated with an even higher degree of precision. Moreover, in contrast to the second embodiment, the motor speed remains the same during the operation performed without applying the test signal and the operation performed by applying the test signal, and therefore the cause of an estimation error in the angle error due to a difference in the operating speed can be eliminated. [0119] Further, the angle error may be estimated a plurality of times using the procedures described above with respect to a plurality of test signals having different initial phases and amplitude values, and an average value thereof may be set as the estimated angle error value. Moreover, this embodiment can be expanded easily to a case in which the angle error includes a plurality of frequency components. [012 0] FourLh Embodiment In a fourth embodiment of this invention, the operations of the angle error estimation unit 52 differ from the first embodiment. Processing executed by the angle error estimation unit 52 according to the fourth embodiment of this invention will be described below with reference to a flowchart shown in Fig. 10. [0121] Note that in Fig. 10, flows marked with identical reference symbols to Fig. 7 denote identical operations to the first embodiment, and description thereof has been omitted. In the fourth embodiment of this invention, operations performed in step S31 and step S33, in which test operations are performed by applying test signals, and step S35, in which the estimated angle error value is calculated, differ from the first embodiment. [0122] In the fourth embodiment of this invention, an operation is performed by generating two test signals in which at least one of the amplitude and the initial phase is different in each test signal, whereupon the angle error is estimated using frequency analysis results obtained in relation to current pulsation components corresponding to the test signals. The frequencies of the test signals are set to be identical to the angle error frequency. [0123] First, in step S31, the motor 2 is operated by applying a first test signal having a predetermined amplitude, a predetermined initial phase, and an identical frequency to the angle error frequency. Next, in step S2, an identical operation to the first embodiment is performed. [0124] Next, in step S33, a similar operation to step 531 is performed, except that a second test signal is set to differ from the first test signal applied in step S31 in terms of at least one of the amplitude and the initial phase of the applied test signal. Next, in step S4, an identical operation to the first embodiment is performed. [0125] Next, in step S35, the angle error is estimated using frequency analysis results obtained during the two test operations implemented in the preceding steps. Processing executed by the angle error estimation unit 52 in step S35 of Fig. 10 to calculate the estimated angle error value will now be described in detail with reference to a flowchart shown in Fig. 11. [0126] Note that the frequency analysis results determined in step S2 and step S4 are frequency analysis results obtained in relation to current pulsation generated in response to a combined signal combining the angle error and the test signal. First, a calculation is performed to separate the current pulsation component derived from the test signal from these frequency analysis results (step S41) . [0127] As an example of this calculation, the current pulsation component derived from the test signal can be separated by subtracting the frequency analysis result obtained in step S2 from the frequency analysis result obtained in step S4, for example. In other words, the difference between the Fourier coefficients determined in step S4 and step S2 serves as the Fourier coefficient of the current pulsation component derived from the test signal. [0128] Huwevei, alLenLion must be paid Lo Lhe fact. LhuL the test signal at this time is a signal (for convenience, referred to as a combined test signal) obtained by performing subtraction on the test signals of step S2 and step S4. In other words, the two test signals are known, and therefore the combined test signals obtained by performing subtraction on the test signals are also known, meaning that the amplitudes and the initial phases thereof are known. [0129] Next, the transfer characteristic at the frequency of the angle error (which is equal to the frequency of the combined test signal) is determined from a relationship between the current pulsation component derived from the combined test signal extracted in step S41 and the combined test signal by implementing similar procedures to the first embodiment (step S42) . As a result, an amplitude and a phase corresponding to Ait and (j>it of the first embodiment can be determined using the extracted combined test signal and the Fourier coefficient of the current pulsation component derived from the combined test signal. [0130] Next, the current pulsation component derived from the angle error is extracted from the frequency analysis result obtained during the test operation (step S43). Here, the test operation may be either the test operation of step S31 or the test operation of step S33. A case in which the test operation result of step S31 is used will be described below. [0131] The first test signal used during the test operation of step 331 is known, and therefore the amp 11 Lu.de arid Lhe phase of the current pulsation derived from the first test signal can be determined using the transfer characteristic determined in step S42. Further, the current pulsation component derived from the angle error is extracted by subtracting the amplitude and the phase of the current pulsation derived from the first test signal from the frequency analysis result determined in step S2 in relation to the current pulsation generated during the test operation. [0132] Next, the estimated angle error value is determined from the current pulsation component derived from the angle error, extracted in step S43, and the transfer characteristic at the frequency of the angle error, determined in step S42, using a similar method to the method described in the first embodiment (step S44) . [0133] In other words, values corresponding to the amplitude An and the phase §±1 of the first embodiment are determined from the current pulsation component derived from the angle error, extracted in step S43. Further, the amplitude and the phase of the angle error are determined from Equations (6) and (7) using the amplitude and phase determined in step S42 corresponding to AlL and ;t of the first embodiment. [0134] In the fourth embodiment of this invention, as described above, the motor 2 is operated in a condition where the frequencies of the test signals are aligned with the angle error frequency, and the angle error is estimated from the corresponding current pulsation component. At this time, the transfer characteristics determined during the operations performed by applying the test signals can be aligned precisely with the angle error frequency, and therefore the angle error can be estimated with an even higher degree of precision. Moreover, in contrast to the second embodiment, the motor speed remains the same during the two operations performed by applying the test signals, and therefore the cause of an estimation error in the angle error due to a difference in the operating speed can be eliminated. [0135] Further, the angle error may be estimated a plurality of times using the procedures described above with respect to a plurality of test signals having different initial phases and amplitude values, and an average value thereof may be set as the estimated angle error value. Moreover, this embodiment can be expanded easily to a case in which the angle error includes a plurality of frequency components. [0136] Note that instead of extracting the current pulsation component derived from the angle error in step S43, the motor 2 may be operated in a condition where the test signals are not applied, and the current pulsation component derived from the angle error may be determined by performing frequency analysis on the motor current at this time. [0137] Fifth Embodiment Fig. 12 is a block diagram showing an overall configuration of a motor control system including an angle error correction device for a position detector according to a fifth embodiment of this invention. In Fig. 12, elements marked with identical reference symbols Lo Fiy. 1 denote identical operations Lo the operations described in the first embodiment. [0138] In the fifth embodiment of this invention, an angle error correction device 5A is provided in place of the angle error correction device 5 shown in Fig. 1. The angle error correction device 5A includes the frequency analysis unit 51, an angle error estimation unit 52A, the angle error correction unit 53, and a resonance determination unit 54A. In other words, the angle error estimation unit 52A performs different operations to the angle error estimation unit 52 shown in Fig. 1, and the resonance determination unit 54A is provided additionally. [0139] The resonance determination unit 54A determines, on the basis of the frequency analysis result obtained by the frequency analysis unit 51 or the estimated angle error value obtained by the angle error estimation unit 52A, whether or not the angle error frequency of the position detector 3 or the frequency of the test signal matches a resonance frequency of the motor control system, and outputs a determination result to the angle error estimation unit 52A. [0140] When the motor 2 is connected to a load, the motor control, system may, depending on the dynamic characteristic of the load, have a resonance point. When the frequency of the angle error or the frequency of the test signal is close to or matches the frequency of the resonance point (i.e. the resonance frequency) during an operation of the motor 2, the precision with which the angle error is estimated may deteriorate. [0141] Hence, to avoid such a case, the angle error correction device 5A according to the fifth embodiment of this invention is capable of estimating the angle error with stability and a high degree of precision, as will be described below. An operation of the resonance determination unit 54A will be described below. The resonance determination unit 54A operates the motor 2 in order to determine whether or not the angle error frequency or the test signal frequency matches the resonance point before the angle error is estimated in the manner described in the first to fourth embodiments. [0142] Here, in a case where the resonance point does not vary according to the rotation position of the motor 2, for example when the load is a rotary device or the like, the motor 2 is operated while varying the operating speed thereof, and frequency analysis is performed on the motor current. [0143] Further, by performing frequency analysis on the motor current, a determination can be made as to whether or not the frequency of the angle error is in the vicinity of the resonance frequency on the basis of the values obtained in Equations (2) and (3) or Equations (4) and (5) . In the vicinity of the resonance frequency, the amplitude of the current pulsation increases or decreases rapidly, while the phase varies rapidly by close to 180 degrees. [014 4] Hence, a determination is made as to whether or not amounts of variation in the amplitude and phase of the current pulsation, determined by frequency analysis, exceed predetermined values, and when the amounts of variation exceed the predetermined values, the operating speed of the motor 2 on the periphery thereof is determined to be close to the resonance frequency. [014 5] Further, on the basis of the determination result obtained by the resonance determination unit 54A, the angle error estimation unit 52A outputs an operation command to the motor control device 1 to operate the motor 2 under conditions for avoiding the resonance frequency. [014 6] Furthermore, the angle error estimation unit 52A estimates the angle error using the methods described in the first to fourth embodiments. In this case, the operating speed of the motor 2 is modified so as to avoid the resonance frequency. When the operating speed of the motor 2 is modified, the frequency of the angle error and the frequency of the test signal change, and therefore the resonance frequency can be avoided. [0147] Hence, in the fifth embodiment of this invention, the resonance determination unit 54A determines whether or not the angle error frequency or the test signal frequency matches the resonance frequency of the motor control system, and estimates the angle error under conditions in which the angle error frequency and test signal frequency do not match the resonance frequency. As a result, the angle error can be estimated with stability and a high degree of precision. In particular, the resonance frequency can be avoided even when a load is attached, and therefore adjustments can be made with a high degree of precision during installation of the motor control system. [0148] In Ihe fifLh embodiment, an example in which Lhe angle error estimation unit 52A outputs an operation command to operate the motor 2 so as to avoid the resonance frequency was described, but instead, a control unit that advances the sequence of angle error estimation operations, including outputting the operation command for operating the motor 2, may be provided separately to the angle error correction device 5 and the motor control device 1 or as a dedicated control device. [0149] Sixth Embodiment Fig. 13 is a view showing a configuration of an elevator control device according to a sixth embodiment of this invention. Here, Fig. 13 is a view showing a case in which the motor control system including the angle error correction device for a position detector according to the first to fifth embodiments of this invention is applied to an elevator. In Fig. 13, parts marked with identical reference numerals to Fig. 1 or Fig. 12 denote identical operations to the operations described in the first to fifth embodiments. [0150] In Fig. 13, a car 7 and a counter weight 9 of an elevator are connected to each other by a hoisting rope 8 and suspended from a sheave 6 in the manner of a well bucket. The sheave 6 is connected to the motor 2, which serves as a motor for driving the car 7, and the car 7 is raised and lowered by power from the motor 2. [0151] Here, the angle error is estimated during installation of a hoisting machine, for example. More specifically, the angle error is estimated while rotating the hoisting machine by installing the molur 2 in an elevaLox system as Lhe hoisting machine and performing an operation for estimating the angle error in a condition where the rope 8 either is or is not wound around the sheave 6. [0152] At this time, the angle error can be estimated with stability by estimating the angle error only during an interval in which the car 7 travels at a constant speed. Further, to extend the interval in which the car 7 travels at a constant speed, an operation may be performed after setting a travel speed to be lower than a rated speed of the elevator. Moreover, to improve the estimation precision, the travel speed of the elevator may be modified to a travel speed at which the amplitude of the current pulsation increases. Note that there are no limitations on the position of the car 7, and the angle error may be estimated in any position within a hoistway through which the car 7 travels. [0153] Further, to increase the amplitude of the current pulsation in order to improve the estimation precision, an operation may be performed after modifying respective gains of the speed control unit and the position control unit so that the gains increase. In the case of PID control, a proportional gain, an integral gain, and a differential gain correspond to the gains of the control device. [0154] Furthermore, the angle error estimation result is recorded in a storage medium (a non-volatile memory, for example) as an angle error corresponding to a magnetic pole position of the hoisting machine. During a normal operation, the estimated angle error value COrrespundiny Lo the uuLpuL of the posiLion delecloi 3 is read from the storage medium and corrected. Information relating to the angle error recorded in the storage medium may be information that indicates the error amplitude and the phase shift of the angle error and is used to determine the angle error by solving Equation (1), or may be corrected angle information or corrected position information that corresponds to the magnetic pole position of the hoisting machine and is provided in the form of a table or the like. In this case, a method of storing phase information and amplitude information in advance and correcting the angle error by calculation is preferably employed in order to minimize the amount of information. [0155] Note that in the elevator, the dynamic characteristic of the elevator system varies according to the position and the load weight of the car 7, and therefore the transfer characteristic shown in Fig. 6 likewise varies according to the position and load weight of the car 7. Accordingly, when the angle error is estimated by performing operations in which correction signals are applied a plurality of times, the operations are preferably performed under conditions in which the position and load weight of the car remain constant or nearly constant. [0156] Further, in the elevator, the dynamic characteristic of the elevator system varies when specifications such as an ascent/descent length and a rated load capacity vary, but in this invention, the transfer characteristic of the motor control system is determined during an operation performed by applying a test signal, and therefore the angle error can be estimated regardless of the specifications of the elevator. Needless to mention, this invention is not limited to an elevator, and may be used to estimate an angle error in any system in which a load characteristic of a motor varies from moment to moment. [0157] Furthermore, in this invention, the angle error can be estimated simply by performing frequency analysis in a minimum of two patterns, and therefore the angle error can be estimated quickly. Moreover, once estimation has begun, estimation is performed continuously without stopping the motor 2, and therefore the angle error can be estimated quickly. Hence, the angle error can be estimated quickly during a test operation following installation of the elevator, for example, meaning that there is no need to secure time for estimating the angle error. As a result, the amount of time required for adjustments during installation can be reduced. [0158] Next, a case in which the angle error is estimated while varying the position of the car 7 will be described. In a case where the angle error is estimated while operating the car 7 from a lowermost floor to an uppermost floor or from the uppermost floor to the lowermost floor during installation, for example, the angle error can be estimated with a high degree of precision by implementing following procedures. [0159] In the elevator, resonance points derived from an elastic characteristic of the rope 8 exist between the car 7 and the rope 8 and between the counter weight 9 and the rope 8 . Further, the resonance points vary according to the position of the car 7 and the load weight of the car. Therefore, the period of the periodic angle error of the position detector 3 and the frequency of the test signal used to estimate the angle error may match the resonance frequencies of the resonance points. Here, when the frequency of the angle error or the frequency of the test signal matches a resonance frequency of the elevator, the amplitude and phase of the current value used during frequency analysis vary rapidly, leading to instability in the frequency analysis result, and as a result, the precision with which the angle error is estimated deteriorates. [0160] Hence, before estimating the angle error, the car 7 of the elevator is operated from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor, and frequency analysis is performed on the motor current at the frequency corresponding to the angle error. When, at this time, the frequency of the angle error is in the vicinity of the resonance frequency, the amplitude of the corresponding current pulsation increases or decreases rapidly, while the phase thereof varies rapidly by close to 180 degrees. [0161] Hence, a determination is made as to whether or not the amounts of variation in the amplitude and phase of the current pulsation, determined by frequency analysis, exceed predetermined values, and when the amounts of variation exceed the predetermined values, the frequency of the angle error on the periphery thereof is determined to be close to the resonance frequency. Further, on the basis of the determination result, the angle error is estimated in a differenl position Lo the position determined to be close to the resonance frequency. Note that the operating speed during angle error estimation may be modified so as not to approach the resonance frequency. Further, the method described above is not limited to an elevator, and may also be applied to a case in which the resonance frequency varies according to the rotation position of the motor 2. [0162] For example, the angle error may be estimated during installation of the elevator as follows. First, the car 7 of the elevator is operated from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor, frequency analysis is performed on the motor current at the frequency corresponding to the angle error, and the amounts of variation in the amplitude and phase of the current pulsation are calculated. [0163] At this time, the car position is stored together with the amounts of variation in the amplitude and phase of the current pulsation. Next, when the operation from the lowermost floor to the uppermost floor or from the uppermost floor to the lowermost floor is complete, a determination is made as to whether or not the amounts of variation in the amplitude and phase of the current-pulsation exceed the predetermined values, and positions in which the amounts of variation do not exceed the predetermined values are extracted. Next, the car 7 is moved to a position in which the amounts of variation in the amplitude and phase of the current pulsation do not exceed the predetermined values, whereupon angle error estimation is implemented. [0164] When the upeidLiun performed to determine wheLher or not the amounts of variation in the amplitude and phase of the current pulsation exceed the predetermined values is an operation from the lowermost floor to the uppermost floor and the operation for estimating the angle error is implemented in the opposite direction, i.e. from the uppermost floor to the lowermost floor, the angle error can be estimated during a single reciprocating operation, and as a result, the amount of time relating to the angle error estimation can be reduced. [0165] On the other hand, when the operation performed to determine whether or not the amounts of variation in the amplitude and phase of the current pulsation exceed the predetermined values is an operation from the uppermost floor to the lowermost floor, the operation for estimating the angle error may be implemented in the opposite direction, i.e. from the lowermost floor to the uppermost floor. With this estimation method, the angle error can be estimated while avoiding deterioration of the estimation precision due to resonance, and therefore the angle error can be corrected accurately. Moreover, the angle error can be estimated accurately during a single reciprocating operation, and as a result, the amount of time required for adjustments during installation is reduced. [0166] Note that the overall device layout, the roping system, and so on of the elevator are not limited to the example shown ir Fig. 13. For example, this invention may be applied to an elevator employing a 2:1 roping system. Further, the position of the hoisting machine constituted by Lhe inoLui 2, fur example, is nuI limited to the example shown in Fig. 13. Moreover, this inventior may be applied to various types of elevators, such as a machine room-less elevator, a double deck elevator, a one-shaft, multi-cai elevator, and an inclined elevator, for example.