DESCRIPTION Vector Controller for Permanent-Magnet Synchronous Electric Motor Technical Field [0001} The present invention relates to vector controllers for ■j '• permanent-magnet synchronous electric motors. Background Art [0002] Vector control technologies for permanent-magnet synchronous electric motors (hereinafter abbreviated as "electric motor") using an inverter are widely employed in industry.. By separately controlling the magnitude and the phase of inverter output voltage, a current vector in an electric motor is optimally controlled, so that torque of the electric motor is fast and instantaneously controlled. Permanent-magnet synchronous electric motors are known as high-efficiency electric motors in comparison with induction motors because no Energizing current is needed due to establishment of magnetic field by the permanent magnet and no secondary copper loss is generated due to no rotor current. For that reason, application ' of permanent-magnet synchronous electric motors to electric railcars has been investigated in recent years. [0003] Subjects with controllers in applying permanent-magnet synchronous;:electric motors to electric railcars are to realize a stable field-weakening operation up to a high speed range and to j "i achieve a stable transition to a one-pulse mode in which inverter loss can be minimized and voltage applied to the electric motors can be maximized. The one-pulse mode is an operation mode for I! ' inverters, in which an inverter outputs, as its output line voltage, square waves having a positive and a negative maximum rectangular voltages of 120 degree durations each that are repeated one after another with a zero voltage period of 60 degrees therebetween, in one cycle, i.e., 360 degrees. The following method is disclosed in Patent Document 1 as a |L related prior-art. A voltage setting unit is provided that receives a voltage fixing command and a voltage command calculated based on a current command. When the voltage fixing command is input, ^ i the voltage; setting unit outputs a voltage command as a new voltage command by setting its magnitude to a predetermined voltage set value. A magnetic-field-direction (d-axis) current command is then corrected using a magnetic-field-direction (d-axis) current correcting value obtained by taking.a proportional-integral control of the difference between the voltage command calculated from the current command and the new voltage command. A modulation] index for the inverter is then calculated from the ■i voltage command to control the inverter, so that, a field-weakening operation is performed. [0004] Patent Document l: Japan Patent Application LaidOpen No. H09-84399 (see paras. [0023] - [0029]): Disclosure of the Invention [0005] j In Patent Document 1 cited above, however, how to generate the voltage fixing command is not disclosed and the voltage setting unit needs to be provided anew. Moreover, a capacitor voltage always fluctuates, so that a maximum voltage that the inverter can output also fluctuates accordingly. In order to maximize voltage applied to the electric motor according to the method disclosed in Patent Document 1, it is necessary to vary a timing of generating the voltage fixing command and to vary a voltage set value, depending on the fluctuation in the capacitor voltage, which involves taking a complicated control. [0006] Furthermore, a value is used as the field-direction (d-axis) current correcting value that is obtained by taking the proportional-integral control of the deviation between the voltage command calculated based on a current command and the new voltage command whose magnitude is set by the voltage fixing command. Accordingly, when the deviation between the voltage command and the new voltage command is not zero, i.e., during an input to the.: proportional-integral control remains not zero, the control operation continuously accumulates an integration value. For that reason, when the voltage command theoretically calculated based on the [current command cannot be set to a value smaller than the voltage command set anew —for example, when a torque command is excessively large for rotation speed of the electric motor— even though a magnetic-field-dircction current is corrected using the magnetic-field-direction current correcting value, the difference hetween the voltage command and the voltage command set anew caniiot be set to zero and the integration value in the proportional-inte;gral control is continuously accumulated, so that the magnetic-field-direction current correcting value ■ ' excessively increases as time elapses. When the magnetic field-direction current correcting value becomes excessively large, the vector control cannot be normally performed. A complicated operation is therefore required in a practice use, such as limiting the integration value to a value less than an upper limit or resetting the integration value under a spocifiod condit ion. [0007] The present invention is made to solve the above-described problems, and provides! a vector controller for a permanent-magnet synchronous electric motor that can realize with a simplified configuration a stable one-pulse-mode field-weakening control in a high speed range. Means for Solving the Problem [0008] A vector controller for a permanent-magnet synchronous electric motor, according to the present invention, controls an alternating; current from an inverter that drives the permanent-magnet, synchronous electric motor so as to come into coincidence with a current command, and provided with a reference phase-angle calculation unit for generating a reference phase angle of the permanent-magnet synchronous electric motor; a current command generation unit. for. generating the current command using a given torque command; a current control unit for making a control calculation of a current error between the current command and a current through the permanent-magnet synchronous electric motor, to output the calculated current error; a decoupling voltage calculation unit for calculating a feed-forward voltage using motor parameters of the permanent-magnet synchronous electric motor and the current command; a modulation index calculation unit for outputting, a modulation index for the inverter by receiving a direct-current voltage to the inverter and a voltage command that is the sum of the current error and the feed-forward voltage; a control phase-angle calculation unit for outputting a control phase angle for the inverter by receiving the voltage^ command and the reference phase angle; a pulse-width-modulation signal generation unit for generating pulse-width-modulation signals for the inverter using the modulation index and the control phase angle; and a Effects of thejlnvention [0009] A vector controller for a permanent-magnet synchronous electric motor, according to the invention, controls an alternating current from an inverter that drives the permanent-magnet synchronous electric motor so as to come into coincidence with a current command, and provided with a reference phase angle calculation unit for generating a reference phase angle of the permanent-magnet synchronous electric motor.' a current command generation unit for generating the current command using a given torque command; a current control unit for making a control calculation of a current error between the current command and a current through the permanent-magnet synchronous electric motor, to output the calculated current error; a decoupling voltage calculation unit for calculating a feed-forward voltage using motor parameters of the permanent-magnet synchronous electric motor and the current command; a modulation index calculation unit for outputting a modulation index for the inverter by receiving a direct-current voltage to the inverter and a voltage command that is the sum of-the current error and the feed-forward voltage! a control phase-angle calculation unit for outputting a control phase angle for the inverter by receiving the voltage command and the reference phase angle; a pulse-widtlrmodulation signal generation unit for generating pulse-width-modulation signals for the inverter using the modulation index and the control phase angle; and a. current command compensator for correcting the current command using a corrected current command calculated based on the modulation index wherein the current command compensator sets the corrected current command to a va,luc obtained by processing through a time delay element and bv multiplying bv a predetermined constant the difference between the modulation index and a predetermined modulation index set value. Therefore, an effect is brought about that can realize with a simplified configuration a stable one-pulse-mode field-weakening control in a high speed range. BrieT Description of the Drawings [0010] I! Fig. 1 is a block diagram illustrating an example of a configuration of a vector controller for a permanent-magnet i synchronous'electric motor, according to Embodiment 1 of the present invention; ., Fig. 2 is a block diagram illustrating an example of a configuration of a current command generation unit in Embodiment 1 of the invention; Fig. 3 ,:'is a block diagram illustrating an example of a j configuration of a PWM-signal generation unit in Embodiment 1 of the invention; i Fig. 4 shows charts-for illustrating a modulation index PMF, pulse-mode transitions, switching operations, and a control-mode transition, with inverter angular frequency o>, in Embodiment 1 of the invention.' Fig. 5 is a block diagram illustrating an example of a configuration of a current command compensator in Embodiment 1 of the invention; Fig. 6 is a graph showing a relation of the deviation between the sum of squares of dq-axis current, commands and that of squares of dq-axis currents, to a d-axis current error, in Embodiment 1 of the invention; •• i: I Fig. 7 is a graph showing a relation of the deviation between the magnitude of current command vectors and.;.that of current i vectors, to the d-axis current error, in Embodiment 1 of the invention; i Fig. 8 is a graph showing a relation of the deviation between the sum of squares of dq-axis current commands, and that of squares of dq-axis current commands, to a q-axis current error, in i Embodiment 1 of the invention; i Fig. 9 is a graph showing a relation of the deviation between the magnitude of current command vectors and that of current vectors, to :the q-axis current error, in Embodiment 1 of the invention! ' , Fig. 10 illustrates charts showing simulated operating waveforms of torque commands, torques, d-axis current commands, d-axis currents, q-axis current commands, and 'q-axis currents, in Embodiment 1 of the invention; and Fig. 11 illustrates charts showing simulated operating waveforms of modulation indexs, corrected current commands, U-phase voltage commands, synchronous three-pulse .FWM mode flags, synchronous one-pulse mode flags, and U-phaso currents, in Embodiment 1 of the invention. Reference Numerals [0011] l: capacitor, i! T- inverter, 3, 4, 5: current sensor, ft: electric motor, !'■ resolver, 8: voltage sensor, 1(): current command generation unit, 11: d-axis fundamental current command generation unit, 14 ■ adder, 15: q-axis current command generation unit, 20: d-axis current control unit, 21: q-axis decoupling calculation unit, (decoupling calculation unit), 22: d-axis decoupling calculation unit (decoupling calculation unit), l! 23: q-axis current control unit, 30: modulation index calculation unit, 40: control phase-angle calculation unit;, 50: PWM signal generation unit, 53: multiplier, if 54: gain adjustment table, 55: voltage command calculation unit, 57: multi-pulse cnrrier'signal generation unit, ! I 58: synchronous three-pulse carrier-signal generation unit, 59: switch, 60: pulse-mode switching process \init, 61, 62, 63: comparator, 1 64, 65, 66: NOT-circuit, 70: inverter angular-frequency calculation unit, 80: current command compensator, 81: limiter, 82: first-order delay element, 83: proportional gain element, 85: parameter error correction unit, 90: three-phase to dq-axis coordinate transform unit, 95: reference phase-angle calculation unit, and 100: vector controller. Best Mode for Carrying Out the Invention [0012] ' Embodiment 1 Fig. 1 is a block diagram illustrating an example of a configuration of a vector controller for a permanent-magnet synchronous motor, according to Embodiment 1 of the present ; i' invention. As shown in Fig. 1, a main circuit is configured with a capacitor 1 that is a direct-current, power-source, an inverter 2 that J converts direct-current voltage of the capacitor 1 into an alternating-current voltage of any given frequency, and a permanent-magnet synchronous electric motor 6 (hereinafter, simply referred to as "electric motor"). The main circuit is also provided with a voltage sensor 8 that senses voltage of the capacitor 1, current sensors 3, 4, and 5 that sense currents iu, iv, and iw through the output lines of the i-i inverter 2. The electric motor 6 is provided with a resolver 7 that senses a rotor mechanical angle 0m. Each of these sensing signals ii is input into' a vector controller 100. i' [0013] ;j The resolver 7 may be substituted with an encoder, or a position sensorless method may be used in which a position signal is calculated from a sensed voltage, current, or the like instead of a position signal obtained by the resolver 7. In these cases, the re.so.lver 7 is 'unnecessary, in other words, acquisition of a position signal is not limited to using the resolver 7. As for the current sensors 3, 4, and 5, a configuration may be employed in which sensors are provided for a I, least two phase-lines since a current through the other phase fine can bo determined by calculation, or respective currents are determined by simulating output currents of the inverter 2 from a current on the direct-current side thereof. [0014] The inverter 2 receives gate signals U, V, W, and X, Y, .Z'thal; :i are generated by the vector controller 100 to take a pulse-width-modulation (PWM) control of switching elements built in the inverter 2. A PWM voltage-source inverter is suitable for the inverter 2, and since its configuration is publicly known, a detailed description thereof is omitted. ' [0015] The vector controller 100 receives a torque command T*from a externa] controller, not shown, and controls the inverter 2 so that torque T produced by the electric motor 6 comes into coincident with the torque command T*. [0016] Next, a .configuration of the vector controller': 100 is described. The vector | controller 100 is configured with a reference phase-angle calculation unit 95 that calculates a reference phase angle 9e from the rotor mechanical angle 0in\ a three-phase to dq-axis coordinate transform unit 90 that generates a d-axis current /(/and a q-axis current iq from the three phase currents iu, iv, and iw sensed by the current sensors 3, 4, and 5, respectively, and from the reference phase angle 6e\ an inverter angular-frequency calculation unit 70 that calculates an inverter angular frequency to from the reference phase angle Oe\ a current command generation unit 10 that generates a d-axis current command id* and a q-axis current command iq* from the torque command T* input from externally and a later-described corrected i current command dV*~, a d-axis current control unit 20 that generates :a d-axis current error pdo by talcing a proportional-integral control of'the difference betwoen the d*axis current command id* and the d-axis current id, a q-axis current control unit 23 that generates a q-axis current error pqe by taking a proportional-integral control of the difference between the q-axis current command j'g*and the q-axis current iq', a q-axis decoupling calculation unit 21 that calculates a q-axis feedforward voltage VqFF from !the d-axis current command id* and the inverter angular frequency a)', a d_axis decoupling calculation unit 22 that calculates a* d-axis feed-forward voltage VdFF from the q-axis current command iq* and the inverter angular frequency &>> a modulation index calculation unit 30 that calculates a modulation index PMF a control phase-angle calculation unit 40 that calculates a control phase angle 0 from a d-axis voltage command Frf*that is the sum of the d-axis current error pde and the d-axis i feed-forward "voltage VdFF, a q-axis voltage command Kg* that is the sum of the q-axis current error pqe and the q-axis feed-forward voltage VqFF, the reference phase angle 9e, and from a later-described control phase-angle correcting value dTHV, a PWM-signal generation unit 50 that generates the gate signals U, V, W, and X, Y, Zfor the inverter 2; a current command compensator 80 for calculating the corrected current command dVhy receiving the modulation index PMF> and a parameter-error correction unit 85 that calculates the control phase-angle correcting value dTHV V >'. I from the d-axis current id, the q-axis current iq, the d-axis current command id*, and the q-axis current command iq*. [0017] i i * Hero, the modulation index calculation unit 30 receives the i d-axis voltage command Vd* that is the Bum of the d-axis current error pde and the d-axis feed-forward voltage VdFF , the q-axis voltage command Vq* that is the sum of the q-axis current error pqe and the q-axis feed-forward voltage VqFF, the reference phase angle Be, and a voltage EFC of the capacitor 1. The PWM-signal generation unit 50 receives the modulation index PMF and the control phase angle O. rooisj Next, detailed configurations of each of the: control blocks mentioned above will be described. The reference phase-angle calculation unit 95 calculates from the t'otor mechanical angle Om ' the reference,phase-angle Be that is an electric angle, based on the ,i following equation (l)' &e= Bm * PP (1), where PP denotes a pole pair number of the electric motor 6, [0019] The three-phase to dq-axis coordinate transform unit 90 generates the d-axis current id and q-axis current iq. from the three phase currents Su, iv, and iw and the reference phase-angle Be, based on the'following equation (2): [0020] the inverter angular frequency a by differentiiiting the reference phase angle Be, based on the following equation (3): ro= dOe! dt (3). [0022] f A configuration of the current command generation unit 10 is ■ described. Fig. 2 is a block diagram illustrating an example of a configuration of the current command generation unit, 10 in Embodiment; 1 of the invention. A d-axis fundamental current command generation unit 11 receives the torque command T*, to generate a d-axis fundamental current command id J*. , A maximum torque control method that can generate a desired torque of the electric motor 6 with a minimum current is known as a method of generating the d-axis fundamental current command idJ*, in which an optimum d-axis fundamental current command idl" is obtained by referencing a map on the basis of the torque command T*or by using an arithmetic expression. Since the unit can be configured by using known art, a detailed description of the unit is omitted here. i [0023] ! ' ■ After the d-axis fundamental current command idl* is generated, the d-axis current command id* is then obtained by ! T ll adding the corrected current command dV to the' d'axis fundamental current command idl*hy an adder 14. The corrected current command dV is provided for so-called fie Id-weakening control. The correcting value dV has a negative value to correct the value idl* \\\ the negative direction, so that the d-axis current command /(/''"increases in the negative direction; whereby magnetic field is generated in such a direction as to cancel magnetic field produced by the permanent magnet of the electric motor 6, so that I flux linkage! of the electric motor 6 is weakened, that is, a field-weakening control is realized. Since a method of generating the corrected; current command dV\% a key feature of the invention, it will be described later. [00211 The q-axis current command iq*is finally generated by a q-axis i current command generation unit 15 from the d-axis current command id*and the command torque valo T*. As for a method of generating the q-axis current command, there has also been a method in which an optimum q-axis current command iq* is obtained by referencing a map or by using an arithmetic expression, as mentioned above. Since the generation unit can be configured by using known art, a detailed description of the unit is omitted here. [0025] The q-axis current control unit 23 and the d-axis current control unit 20 generate the q-axis current error pqe and the d-axis current error pde obtained by proportional-integral amplification of the differences between the q-axis current command iq* and the q-nxis. current, and between the d-axis current command id* and the d-axis current, based on the following equations (4) and (5), respectively:' i pqe = ( Ki+ Kg I s) * ( iq*- iq) (4) and pde =(K3+ K4I s) * ( id*- id) (5), where Ki and K.i are proportional gains, and K% and K.t are integral gains. i As described later, the q-axis current error pqe and the d-axis current error pde are gradually decreased to zero after a transition from a control mode 1 (described later) to a control mode 2 (described later), and are gradually increased when transition is made from the control mode 2 to the control mode 1 on the contrary. [0026] The d-axis decoupling calculation unit 22 and the q-axis i decoupling calculation unit 21 that are decoupling voltage calculation units calculate the d-axis food-forward voltage VdFF and the q-axis feed-forward voltage VqFF, based on the following equations (6) and (7), respectively: VdFF= ( R,+ s * Ld) * id*- n ) (7), where Ri denotes primary winding resistance (fl), Ld nnd Lq denote fi d-axis inductance (H) and q-axis inductance (H), respectively, *is low, a modulation index PMF is small and the switch 59 is selected to the A side, i.e., a pulse mode is set to the multi-pulse PWM mode. At the same time, a control mode is set to the control mode 1, and the q-axis currenl; control unit 23 and the d-axis current control unit 20 operate according to the above equations (4) and (5), respectively. When the modulation index PMF equals to or exceeds 0.785 with increasing electric railcar speed, since an output voltage of the inverter 2 is saturated in the multi-pulse PWM mode, the switch 59 is changed to the B side to set the pulso-mode to the synchronous three-pulse PWM mode. [0039] At the same time, the control mode is selected to the control mode 2, and the d-axis current control unit 20 mid the q-axis current control unit 23 cease their calculations so that their outputs ar.e reduced to zero. The reason for reducing to zero is as follows. In the synchronous three-pulse PWM mode, the number of pulses per half cycle of output voltage of the inverter decreases from ten or more in the multi-pulse PWM mode to three, so that a control delay increases. If the d-axis current control unit 20 and the q-axis current control unit 23 are left to continue the calculations, there is a fear that the control system becomes unstable. Therefore, the d-axis current control unit 20 and the q-axis current control unit 23 are stopped . to make their calculations. In addition, it is preferable for avoiding a shock produced at the mode change that the outputs of the d-axis current control unit 20 and theqaxis current control unit 23 be gradually reduced i toward zero; with a predetermined time constant during the ,i reduction process. [0040] In control mode 2, a mismatch between electric motor parameters and control parameters arises from stopping the calculations of the d-axis current control unit 20 and the q-axis' ii current control unit 23. Deviations of a torque of and a current through the electric motor 6 from their command values are generated from the mismatch and like. Control errors such as the deviations can be suppressed by correcting the control phase angle fusing the control phase-angle correcting value dTHV generated by the parameter-error correction unit 85 using the d-axis current id, the q-axis.current iq, the d-axis current command id*, and the q-axis current command iq*. A detailed configuration of the parameter-error correction unit 85 will be described later. in addition, the output of the paranietor-m-ror correction unit 85 ia increased after the change from the control mode 1 to the ■i control mode 2, and decreased to zero after the change from the control mode 2 to the control mode 1 on the contrary. The increase and the decrease are preferably performed slowly with a predetermined time constant. An unstable control can thereby be avoided that results from competition of the output of the d-axis current control unit 20 or the q-axis current control unit 23 with that of the parameter-error correction unit 85. [0042] When the modulation index PMF becomes 1.0 or larger with electric railcar speed being further increased, the pulse mode is changed to the one-pulse mode by changing the switch 59 to the C side. The control mode still remains the control mode 2. When the electric railcars slow down by regenerative brakes, which is not shown, the pulse mode is transited from the one-pulse mode to the multi-pulse PWM mode through the synchronous three-pulse PWM mode and the switch 59 is changed from the C side to the A side through the B side in a sequence reverse to the above, so that the control mode; is transited from the control mode 2 to the control mode 1. ;| [0043] Next, a description is made of a configuration of the current command compensator 80 that is a key component to demonstrate effects of the invention. Fig. 5 is a block diagram illustrating an example of a configuration of the current command compensator 80 in Embodiment 1 of the invention. As shown in Fig. 5, the difference between a modulation index net value PMFmax and a i modulation index PMF\s input into a limiter 81 that is able to limit i tho difference to the range between an upper limit and a lower limit. The limiter 81 is configured so as to be able to limit its input signal to the range between an upper deviation-limit set; value LIMHand a lower deviation-limit set value LIML, to output; the limited signal. i The output of tho limiter 81 is input into a first-order delay element 82. An output of the first-order delay element 82 is input into a proportional gain element 83 and multiplied by a gain Jf that is a predetermined coefficient, to be output as the corrected current command dV. With the first-order delay element 82, even when the difference between the modulation index set value PMFmax and the modulation index PMF upsurges, the corrected current command rfKincreases with a predetermined time constant. As described above, the corrected current command rfFis expressed as the following equation (16): dV= LIMHL ( PMFmax - PMF) * ( 1 f ( 1 + s r) ) * K (16), where LIMHLi ) denotes a function that limits a value in the parenthesis to the range between the upper deviation-limit set value LIMH and the lower deviation-limit set value LIML, and r denotes a first-order delay time constant. The time constant r is 10 ins to 100 ms orders of magnitude. [004 4] Preferable settings of the modulation index set value PMFmnx, the upper deviation-limit set value LIMB, and the lower deviation-limit set value LIML are as follows for Embodiment. The modulation index set value PMFmnx is preferably set to 1.0. This is because that, at. the time when a modulation inde* PMF reaches 1.0, i.e., an output voltage of the inverter 2 reaches its maximum voltage, an input to the limiter 81 becomes zero or less, so that, a negative corrected current command dVc-An be generated, ■h which is preferable for performing a field-weakened control while the output voltage of the inverter 2 is maxim.iy.ed. [0045] The upper deviation-limit set value LIMHia preferably set to a value that is obtained by dividing by the gain K the maximum i d-axis current Id max (referred to as "maxim um field-wen kening current"), which is calculated in advance, required to flow through the electric motor 6 when producing a desired torque command T*, taking into account a fluctuation range of the capacitor voltage EFC. For example, when the maximum d-axis current Idmax is 100 A and the gain K is set to 100,000, the upper deviation-limit set .L I value LIMIT becomes 0.001, The lower devintion-limit set value LIML is preferably set to zero. By thus setting the set values, when a modulation index PMFis 1.0 or smaller, i.e., when there is a margin between a voltage command and the maximum output voltage of the inverter 2, the corrected current command dVis not output. At the time when the modulation index PMFexceeds 1.0, i ;. i i.e., when a voltage command slightly exceeds the maximum output voltage of the inverter 2, the limiter 81 generates a negative output value, so that a corrected current command dV is output. An unnecessary d-axis current id therefore does not flow, which allows a current through the electric motor 6 to be minimized. [0046] ; By thus generating a corrected current command rfFbased on * ii- the modulation index PMF thnt is a value obtained by normalizing the magnitude of a command output-voltage vector for the inverter by the voltage EFC of the capacitor 1, an appropriate corrected current command dV can be obtained, independently of the magnitude of the voltage EFC of the capacitor 1, depending on an excess ratio of the command output-voltage vector magnitude for the inverter to the maximum voltage thai; the inverter 2 is able to output. Accordingly, stable operation can also be obtained in application to electric railcars whose voltage EFCoi the capacitor 1 fluctuates. [0047] Furthermore, by generating the corrected current command dV using (".lie combination of the proportion;)! gain uleinenl. 83 and the; first-order delay element 82, a stable; operation can be performed even when the electric motor 6 falls into an operation range where; the field-weakening control is not theoretically applied, for example, when a torque command T*is excessive for rotation speed of the electric motor 6. In such a situation, even though the d-'axis current command id* is corrected to a negative by a corrected current, command dV, the magnitude of the command Output-voltage vector for the inverter cannot be reduced to the maximum voltage or. less that the inverter is able to output. Namely, with the combination of the proportional gain element 83 and the first-order delay element 82, even in the situation where a modulation index PMFremains more than 1.0, a final value of the corrected current command dV never continue to increase to an ■i excessively large value in the configuration of the invention, since the correcting value rfKsettles to an appropriate value determined from the modulation index PMF, the upper deviation-limit set value LIMH, and the gain K. In other words, even when the torque command T* is excessive, an appropriate field-weakening control can be performed. [0048] In a case of a configuration made with a proportional-integral controller having an integral element ns seen in conventional ■i configuration examples, instead of the above-mentioned combination of the gain K and the first-order delay element 82, when a modulation index PMF remains larger than 1.0, an integration value is accumulated in the integral element and a corrected current command dV continues to increase to an excessively large value as time elapses, so that I.he electric motor 6 cannot bo properly controlled. Moreover, oven when tho motor recovers from such an uncontrolled state to tlie normul state, it. takes time to decrease the excessively accumulated integration value to a proper value, which brings poor control during this interim period. For that reason, a complicated operation is ■i required in a'practice use, such as setting of an upper limit for the integration value or resetting of the integration value at a predetermined timing. According to the invention, on the contrary, no such complicated „ operation is needed to perform a stable field "weakening control. [0049] 'l Next, a description is made of a configuration of the parameter-error correction unit 85 that is a key component to demonstrate the effects of the invention. The parameter-error correction unit 85 calculates the control phase-angle correcting value dTHV from the d-axis current id, the q-axis current iq, the d-axis current command id*, and the q-axis current command iq*, based on the following equation (17)'- dTHV = (Kg+Kef s)*{ ( id** + iq**) - (id* + iq2)) (17), where Ks and Kg denote a proportional gain and an integral gain, respectively,* whereby the correction unit operates as a I! proportional-integral controller. The first term on the right-hand side of the equation (17) expresses the sum of squares of the d-axis current command j'tf *and the q-axis current command iq*, and denotes the square of the magnitude of the current command vector. The second term on the right-hand side expresses the sum of squares of the d-axis current id and the q-axis current iq, and denotes the square of the magnitude of the current vector. [00501 The daxis current /tf and the q-axis current /tfthat are electric motor . currents may sometimes deviate from the d-axis current ■< command id* and the q-axis current command iq* that are current commands, respectively, by permanent-ma gno I; flux 0„ variation and electric-motor parameter variation due to tempornture rise of and current through the electric motor. In this CM He, by subtracting the square of the current vector magnitude from that of the current command vector magnitude and by taking a proportional-integral control of the subtraction result, the control phase angle 8 is corrected using the control phase-angle correcting value dTHV corresponding to the deviation. An operation can thereby be performed so that the electric motor current comes into coincidence with the current command, which allows preventing the torque T of.the electric motor 6 from deviating from the torque command ^.therefor. % [0051] Instead iof using the equation (17), the control phase-angle correcting value dTHVmny be calculated using an equation (18): dTHV= ( K5 + Kef s) * ( SQRTC id** + iq*2) - SQRT( id* + iq2) ) ; (18). The first term on .the right-hand side of the equation (18) expresses the square root of the sum of squares of the d-axis current command id* and the q-axis current command iq*, and denotes the magnitude of the current command vector. The second term on the right-hand side expresses the square root of the sum of squares of She d-axis current; id and the q-axis current iq, and denotes the magnitude of the current vector. [0052] The d-axis current id s\nd the q-axis current iq that are electric i motor currents may sometimes deviate from the d-axis current command id*and the q*axis current command iq* thnt are current commands, respectively, by permanent-magnet; flux $n variation ■i i and electric-motor parameter variation due to temperature rise of i and current through the electric 'motor. In this case, by subtracting the current vector magnitude from the current comminu'l vector magnitude and by tnlcing a proportional-integral conl.rol of the aubtraction result, the conl-rol phase angle Q is corrected using the control phase-angle correcting vuluo ci'J'HV corresponding to the deviation., An operation can thereby be performed so that the electric motor current comes into coincidence with the current command, which allows preventing the torque ^of the electric motor 6 from deviating from the torque command T* therefor. [0053] In addition, since the equation (18) is a complicated equation owing to including two square-root operations in comparison with the equation (17), the calculation takes time and involves a significant load on a microcomputer. Accordingly, it is preferable to use the equation (17). [0054] The difference between a control phase-angle correcting value dTHVcalculated using the equation (17) and that calculated using the equation (18) is explained below. Fig. 6 is a graph showing a relation (obtained using the equation (17)) of the deviation between the sum of squares of dq-axis current commands, and that of squares of dq-axis currents to the d'axis current error, in Embodiment 1 of the invention. Fig. 7 is a graph showing a 'r i, relation (obtained using the equation (18)) of the deviation between the magnitude of the current, command vector and (-hut of the current vector to the d-axis current, error, in Embodiment 1 of the i invention. Fig. 8 is a graph showing a relation (obtained using the equation (17)) of the deviation between the sum of squares of dq-axis current commands, and that of squares of dq-axis currents to the q-flxis'current error, in Embodiment 1 of I;he invention. Fig. 9 is a graph showing a relation (obtained using the equation (18)) of the deviation between the magnitude of the cvtrrenl. command vector and that of the current vector to Lhe q-axis current error, in Embodiment 1 of the invention. [00 55 J In Figs.' 6 and 7, respectively shown are a relation of the deviation (vertical axis) of the sum of squares of dq-axis currents from that of squares of dq-axis current commands, and a relation of the deviation (vertical axis) of the current-vector magnitude from the current command vector magnitude, to a d-axis current error Aid (horizontal axis), when there is an error between the d-axis current id and the d-axis current command id* in a situation of the q-axis current iq being equal to the q-axis current command iq*, i.e., in a situation of the q-axis current error being zero. Here, the d-axis current error Aid denotes the subtraction of the d-axis current id (vom the d-axis current command id*. As shown in Figs. 6 and 7, it is found that both deviations have a similar characteristic such that they are substantially linear to the d'Hxis current error Aid in a range thereof being small (within ±50 A) although the vertical scales are different. In addition, the ■j difference of; the vertical scales is insignificant since it can be adjusted by the gain K5 in the equation (17). [0056] In Figs.. 8 and 9, respectively shown are a relation of the deviation (vertical axis) of the sum of squares of dq-axis currents from that of squares of dq-axis current commands, and a relation of the deviation (vertical axis) of the current vector magnitude from the current command vector magnitude, to a q-axis current error Aiq (horizontal axis), when there is an error between the q-axis current iq and the q-axis current command iq* in a situation of the d-uxis current /(/being equal to the d-axis current command id*, i.e., in a situation of the d-axis current error being zero. Here, the q-axis current error Aiq denotes the subtraction of the q-axis current command iq* h-om the d-axis current iq. As shown in Figs. 8 and 9, it is found thai; both deviations have a similar characteristic such that the deviations are substantially linear to the q-axis current error Aiq in a range thereof being small (within ±50 A) although the vertical scales are different. In addition, the difference of the vertical scales is insignificant since it can be adjusted by the gain K5in the equation (17). [0057] As described above, by using the equation (17), (.he control phase-angle correcting value dTHV can be calculated without; lengthening the calculation time nor involving a significant load on the microcomputer. [0058] Figs. 10 and 11 illustrate charts showing simulated operating waveforms in Embodiment 1 of the invention. In Fig. 10, simulated operating waveforms of torque commands, torques, d-axis current commands, d-axis currents, q-axis current commands, and q-axis currents are illustrated, and in Fig. 11, simulated operating waveforms of modulation indexs, corrected current commands, U-phase voltage commands, synchronous three-pulse PWM mode flags, synchronous one-pulse mode flags, and U-phase currents. As illustrated in Figs. 10 and 11, it is found t that a stable operation is achieved during a power operation (during a time interval of 0 sec to 2.5 sec) and a regenerative operation (during a time interval of 2.7 sec to 5.3 sec), which is i described in detail below. [0059] During the time from 0 sec to around 0.7 sec, a voltage applied to the electric motor 6 and also the modulation index PMF linearly increase, and the multi-pulse PWM mode (its mode flag is not indicated in the figures) and control mode 1 are selected. Since l;he modulation index PMF equals to or exciseds a predetermined value-: at the time around 0.7 sen, the synchronous three-pulse PWM mode and control mode 2 an; selected. .During the time from 0.7 sec to around 1.0 sec;, the modulation index PMF further linearly increases but its magnitude is smaller than 1.0. In addition, the magnitude of the U-phasc voltage command i Vu* decrease's immediately after the change to the synchronous three-pulse PWM mode at the time around 0.7 sec, this is due to the command voltage magnitude PMFM that has been multiplied by 1.274 by the gain adjustment table 54 in the multi-pulse PWM mode is change to be multiplied by 1.0 as described above. [0060] During from the boot-up to the time around 1.0 sec, the maximum torque control is taken by the current command generation unit 10, and the d-axis current command id* and the i ■i q-axis current command iq* are constant because the torque command T* is constant. [0061] At the time around 1.0 sec, since the modulation index PMF reaches 1.0, the synchronous one-pulse mode is selected as the pulse mode and the corrected current commnnd dV negatively increases, so that the d-axis current command id*further increases negatively accordingly. The d-axis current id follows the d-axis current command id*, to negatively increase. From that, it is found that the field-weakening control is desirably performed aw well as the modulation index PMFis kept at a value infinitely close to 1.0, that is, the terminal voltage of the electric motor 6 is kept constant. From the fact that the torque follows the torque command T*. it is found that'the electric motor 6 stably accelerates its rotation speed because the torque command T* is reduced in inverse proportion 1:0 the rotation speed in order that the electric motor G it operates to output constant power. i [00621 At the .time around 1.8 sec, the torque command T* is once reduced to zero to stop the inverter 2 (the gate signals U, V, Wt and X, Y, Z are all switched off). Then, the inverter is rebooted in a power operation mode at the time around 2.0 sec and operated in i the power operation mode till the time around 2.5 sec. It is found that the torque Tis also in coincidence with the torque command T* during the series of such operations, proving that the normal operation.is performed. i Moreover, since the pulse modes are switched over depending on the modulation index PMF, it is found, from the flags of the synchronous three-pulse PWM mode and those of the synchronous one-pulse PWM mode, that the pulse mode is automatically changed to the synchronous three-pulse PWM mode when the modulation index PMF becomes smaller than 1.0 during the processes of the reduction of the torque command T* and of the reboot. [0063] During the time from around 2.2 sec to 2.3i sec, the torque command T* becomes large with respect to the rotation speed, so that the electric motor 6 is operated in a range where the field-weakening control thereof is not theoretically realized. In that range, even though the d-axis current command id* is corrected to a negative using the corrected current command dV, the magnitude of the command output-voltage vector for the inverter cannot be reduced to the maximum voltage or less that the inverter is able to output. Since the corrected current command •i dV is however limited to those within a constant value (-150 A) that is determined from the modulation index PMF, the upper deviation-limit set value LIMHand the gain K, it is found that the corrected current command rfVdoes not become excessively large. [0064] At, the time around 2.7 sec, the torque command T* is sol; to be negative, an;d the inverter is booted in the regenerative operation mode. At the time around 3.2 sec, the torque command 71'*' is once set to zero to stop the inverter 2 (the gate signals If, V, W, and X, Y, Z arc all switched off) and then the inverter is rebooted at; the time around 3.4 sec. It is found that the torque 7'is also in coincidence with the torque command T* during the series of such operations, proving that the normal operation is performed. It is also found that since the modulation index PMF becomes i smaller than 1.0 during the processes of increasing and reducing the torque command T*, the pulse mode is automatically changed to the synchronous three-pulse PWM mode, and that the synchronous one-pulse PWM mode is automatically selected at the stage when the modulation index PMF reaches 1.0. I) ; [0065] While the regenerative operation is continuously performed after the time around 3.4 sec, the d-axis current command id* is adjusted to be negative using the corrected current command dV, so that the field-weakening control is normally performed till the time around 4.2 sec. [0066] After the time around 4.2 sec, since the terminal voltage of the electric motor 6 decreases owing to rotation speed reduction thereof, the modulation index PMFbecomes less than 1.0 and the corrected current command dV automatically becomes zero. At the same ■r i time, the pulse mode is changed to the synchronous three'pulse PWM mode. With further decrease of the modulation index PMF at the time.around 4.5 sec, the pulse mode is changed to-the multi-pulse PWM mode and the control mode 1 is simultaneously selected. [0067.1 In this way, it is found that the stable operations can he performed even in the field-weakening operation range and the transitions between the field-weakening operation range and the other ranges are also stably achieved- It is further found that the transitions between the control modes and between the pulse modes can be stably: achieved. [0068] As described above, the present invention can provide a vector controller for a permanent-magnet synchronous electric motor that can perform stable transitions in operation mode under a range from a low to a high rotation speed of the electric motor 6 with pulse modes and control modes of the inverter 2 being switched over, and can perform, with a configuration more simplified than conventional ones, a stable field-weakening operation in a one-pulse mode in which output voltage of the inverter 2 can be maximized in the high speed range. The pulse modes and the control modes may be switched over » based not on' a modulation index but on a voltage command, motor frequency, inverter frequency, railcar speed, or the like. [0069] The configuration described in Embodiment, is an exemplar of the subject matter of the present invention and can he combined with another'prior art. Modifying the configuration, for example, omitting part thereof can also he made within the scope of the invention [0070] While the subject matter of the invention has been described using an application to a controller for electric railcars in the specification, applicable fields are not limited to this. The invention can be applied to various related fields such as electric vehicles and "elevators. Amended CLAIMS What is claimed is: ;i. A vector controller for a permanent-magnet synchronous electric motor that controls an alternating current from an inverter that; drives the permanent-magnet synchronous electric motor so as to come into coincidence with a current command, comprising: a reference phase-angle calculation unit for generating a reference phase angle of the permanent-magnet synchronous electric motor; a current command generation unit; for generating the current command using a given torque command; a current'control unit for making a control calculation of a current error between the current command and a current through the permaneirt-magnet synchronous electric motor, to output the calculated current, error; a decoupling voltage calculation unit for calculating a feed-forward , voltage using motor parameters of the permanent-magnet synchronous electric motor and the current command; a modulation index calculation unit for outputting a modulation index for the inverter by receiving a direct-current voltage to the inverter and a voltage command that is the sum of the current error and the feed-forward voltage; a control phase-angle calculation unil; for outpul.ting a control phase angle for the inverter by receiving the voltage command and the reference phase angle; a pulse-width-modulation signal generation unit for generating pulsewidthmodulation signals for the inverter using the modulation index and the control phase angle; and a current command compensator for correcting the current command using a corrected current command calculated based on the modulation indexri wherein the current command compensator sets the corrected current command to a value obtained bv processing through a time delay element and bv multiplying bv a predetermined constant the difference between the modulation index and a predetermined modulation index set value. 3. The vector controller for a permanent-magnet synchronous electric motor of claim 1, wherein the modulation index is defined as having a value of unity when the inverter outputs square waves whose fundamental wave component of line voltage of the inverter i reaches a maximum. 4. The vector controller for a permanent-magnet .synchronous electric motor ; of any one of claim 2 through claim 4 claim 1 wherein the modulation index set value in set, to a modulation index by which the inverter outputs square waves whose fundamental i wave component of line voltage of the inverter reaches a maximum. 5. The vector controller for a permanent-magnet synchronous ■i electric motor of claim 4 2_, wherein the upper deviation-limit set value is larger than zero and the lower-deviation-limit set value is equal to or smaller than zero. h i 6. The vector controller for a permanent-magnet synchronous electric motor of claim 4 2., wherein the upper deviation-limit set value is set based on a maximum field-weakening current necessary for the permanent-magnet synchronous electric motor to generate the command torque within a range of variation in the direct-current voltage to the inverter. 7. The vector controller for a permanent-magnet synchronous electric motor of claim 1, wherein the pulse-width-modulation signal generation unit switches over pulse modes of the inverter depending on the modulation index. 8. The vector controller for a permanent-magnet synchronous electric motor of claim 1, wherein the pulse-width-modulation signal generation unit can hold a carrier signal at zero depending i on the modulation index. 9. The vector controller for a permanent-magnet synchronous electric motor of claim 1, further comprising n parameter-error correction unit.for calculating, from the current command and the current through the permanent-magnet synchronous electric motor, a control phase-angle correcting value used for correcting the control phase angle. .10. The vector controller for a permanent-magnet synchronous electric motor of claim -1-t 9_, wherein the control phase-angle correcting value is calculated, by computing the vector control in a i rotating coordinate system having a d-axis and a q-axis orthogonal to each other, based on the sum of squares of a d-axis and a q-axis components of the current command, and on that of squares of a d-axis and a q-axis components of the current through the permanont-niagnet synchronous electric motor. H. The vector controller for a permanent-magnet synchronous electric motor of claim -Hr 9, wherein the parameter"error correction i unit determines whether to make the calculation based on a predetermined signal. 12. The vector controller for a permanent-magnet synchronous electric motor of cJaim -1-3 1±, wherein the predetermined signal is the modulation index. i 13. The vector controller for a permanent-magnet synchronous electric motor of claim J, wherein the current control unit determines whether to make the calculation based on a predetermined signal. 14-. The vector controller for a permanent-magnet synchronous electric motor of claim 4-5• JL3., wherein the predetermined signal is the modulation index. 15. Tho vector controller for a permanent-magnet synchronous electric motor .. of claim 1.0. wherein the con trol phase-angle correcting' value is calculated by talcing ..a. proportional/integral control of, the deference between the,.sum of sjjuarcg.of a fl-axls and a o-axis comnonents of: the current com m and ...and the sum of squares of a, d.-axig and a q-axis components of tho current through the permanent-magnet synchronous electric motor. 16. The vector controller for a permanent-magnet synchronous electric motor of claim 12, wherein the pulse-width-modulation signal generation unit switches over pulse modes of the inverter depending o.n the modulation index, and when the modulation index is_ higher. than that corresponding to a lower limit at which a synchronous three-pulse pulse'width-modulatioa mode is sel.eeted for the inverter, the current control unit does not make its calculation and the parameter-error correction unit makes its calculation instead, and when the modulation index is low_er_than that corresponding to the lower limit at which the synchronous three-pulse pulse-width-modulation mode is selected for the inverter, the current control unit makes its calculation and the parameter-error correction unit does not make its calculation.