1 FORM 2 THE PATENTS ACT, 1970 (39 of 1970) & THE PATENTS RULES, 2003 COMPLETE SPECIFICATION [See section 10, Rule 13] ESTIMATION DEVICE AND AC MOTOR DRIVE DEVICE; MITSUBISHI ELECTRIC CORPORATION, A CORPORATION ORGANISED AND EXISTING UNDER THE LAWS OF JAPAN, WHOSE ADDRESS IS 7-3, MARUNOUCHI 2-CHOME, CHIYODA-KU, TOKYO 1008310, JAPAN THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE INVENTION AND THE MANNER IN WHICH IT IS TO BE PERFORMED 2 DESCRIPTION Field [0001] The present invention relates to an estimation device that estimates at least one of the position and the 5 speed, of an AC motor, and to an AC motor drive device. Background [0002] Information indicating the position of the rotor is used in driving an AC motor such as an induction machine 10 or a synchronous machine. The use of a position sensor or a speed sensor for acquiring information indicating the position of the rotor presents a problem of an increase in manufacturing cost, for example. In view of such a problem, regarding AC motor drive devices, many studies have been 15 made on position sensorless control, which does not use position sensors or speed sensors. [0003] In addition, regarding AC motor drive devices, many studies have been made on reduction of the number of current sensors in order to reduce manufacturing cost, 20 among which one-shunt current detection has been widely used as an inexpensive current detection method. One-shunt current detection is a method for measuring a phase current flowing through the AC motor, using a current sensor provided on the DC bus of the inverter. The reason why the 25 method for measuring a phase current using a current sensor provided on the DC bus is called one-shunt current detection is that the current sensor is often implemented by a shunt resistor. However, a current sensor other than a shunt resistor may be used as the current sensor provided 30 on the DC bus, in which case the method using the other current sensor is still generally called one-shunt current detection. A known example of this type of one-shunt 3 current detection uses a current sensor called a current transformer (CT) different from a shunt resistor. [0004] Patent Literature 1 discloses a technique for driving an AC motor using position sensorless control and 5 one-shunt current detection in combination. In one-shunt current detection, currents of different phases cannot be simultaneously detected at a vertex of the carrier signal. In view of this, the technique described in Patent Literature 1 involves performing interpolation processing 10 on phase currents obtained from the bus current in the second half of a first carrier period and the first half of a second carrier period following the first carrier cycle, thereby calculating phase currents at the vertex of the carrier signal which is the timing of the boundary between 15 the first carrier period and the second carrier period. The technique described in Patent Literature 1 further involves generating a voltage command on the basis of the phase currents at the vertex of the carrier signal, and calculating information indicating the rotation state of 20 the rotor of the AC motor on the basis of the generated voltage command. The information indicating the rotation state of the rotor is at least one of the position and the speed of the rotor. 25 Citation List Patent Literature [0005] Patent Literature 1: Japanese Patent Application Laid-open No. 2015-139359 30 Summary Technical Problem [0006] The technique described in Patent Literature 1 uses the bus current detected in each of the second half of 4 the first carrier period and the first half of the second carrier period following the first carrier period. Depending on the timing when a plurality of switching elements of the inverter switch between on states and off 5 states, it may be difficult to obtain three-phase currents from the bus current in both the first half and the second half of the carrier periods. For the technique described in Patent Literature 1, thus, the period of detection of phase currents for use in calculating the rotation state of 10 the rotor of the AC motor is twice or more a carrier period. Given that, in general, inverter switching loss increases with an increase in carrier frequency, there is generally an upper limit on the carrier frequency in view of the cooling performance or power efficiency of the AC motor 15 drive device. As the rotation frequency of the AC motor approaches the carrier frequency, the control period becomes longer relative to the rotation frequency of the AC motor, which may make it difficult to secure the estimation accuracy of the rotation state of the rotor included in the 20 AC motor. [0007] The present invention has been made in view of the above, and an object thereof is to obtain an estimation device capable of improving the estimation accuracy of at least one of the position and the speed of an AC motor. 25 Solution to Problem [0008] In order to solve the above-described problems and achieve the object, an estimation device of the present invention comprises a bus current detection unit, a phase 30 current determination unit, a time difference calculation unit, and an estimation unit. The bus current detection unit detects a value of a bus current that is a current flowing through a DC bus of a voltage-source inverter 5 driven by a plurality of gate pulse signals generated on a basis of a voltage command. The phase current determination unit performs a determination process for determining values of a plurality of phase currents 5 supplied from the voltage-source inverter to an AC motor on the basis of values of the bus current detected by the bus current detection unit and states of the plurality of gate pulse signals. The time difference calculation unit calculates a time difference between a first current 10 detection time identified as a time of detection of the bus current used in a previous determination process by the phase current determination unit and a second current detection time identified as a time of detection of the bus current used in a present determination process by the 15 phase current determination unit. The estimation unit estimates at least one of a position and a speed, of the AC motor on the basis of the values of the plurality of phase currents determined by the phase current determination unit and the time difference calculated by the time difference 20 calculation unit. Advantageous Effects of Invention [0009] The present invention can achieve the effect of improving the estimation accuracy of at least one of the 25 position and the speed of the AC motor. Brief Description of Drawings [0010] FIG. 1 is a diagram illustrating an exemplary configuration of an electric motor system including an 30 estimation device according to a first embodiment of the present invention. FIG. 2 is a diagram for explaining a method for determining the values of phase currents by a phase current 6 determination unit according to the first embodiment. FIG. 3 is a diagram for explaining a method for determining the values of phase currents by the phase current determination unit according to the first 5 embodiment. FIG. 4 is a diagram for explaining a method for determining the values of phase currents by the phase current determination unit according to the first embodiment. 10 FIG. 5 is a diagram illustrating an exemplary relationship between timings of detection of the bus current for use in phase current determination by the phase current determination unit, the carrier wave, the voltage commands, and the gate pulse signals, according to the 15 first embodiment. FIG. 6 is a diagram illustrating an exemplary relationship between the three-phase voltage commands and the carrier wave during the rotation of the AC motor at low speed, according to the first embodiment. 20 FIG. 7 is a diagram illustrating an exemplary relationship between the three-phase voltage commands and the carrier wave during the rotation of the AC motor at high speed, according to the first embodiment. FIG. 8 is a diagram illustrating exemplary changes in 25 the voltage command of the intermediate phase in the case where the three-phase voltage commands are sine waves, according to the first embodiment. FIG. 9 is a diagram illustrating exemplary changes in the voltage command of the intermediate phase among the 30 three-phase voltage commands modified using third harmonic superposition, according to the first embodiment. FIG. 10 is a diagram illustrating an example of the three-phase voltage commands subjected to zero-vector 7 modulation by a zero-vector modulation unit according to the first embodiment. FIG. 11 is a diagram schematically illustrating integration on the basis of rectangular approximation. 5 FIG. 12 is a diagram schematically illustrating integration on the basis of rectangular approximation. FIG. 13 is a diagram schematically illustrating integration in which fluctuations in time difference are ignored. 10 FIG. 14 is a diagram for explaining an error in differentiation. FIG. 15 is a diagram illustrating an exemplary configuration of an electric motor system including an estimation device according to a second embodiment of the 15 present invention. FIG. 16 is a diagram illustrating an exemplary configuration of an estimation unit according to the second embodiment. FIG. 17 is a diagram illustrating an exemplary 20 configuration of an electric motor system including an estimation device according to a third embodiment of the present invention. FIG. 18 is a diagram illustrating an exemplary configuration of an inter-detection-time voltage 25 computation unit according to the third embodiment. FIG. 19 is a diagram for explaining a method of calculating the three phase voltages between current detection times by the inter-detection-time voltage computation unit according to the third embodiment. 30 FIG. 20 is a diagram illustrating an exemplary result of speed estimation with the time difference between current detection times being a fixed value, according to the third embodiment. 8 FIG. 21 is a diagram illustrating the result of fast Fourier transform (FFT) analysis of the speed estimation result illustrated in FIG. 20. FIG. 22 is a diagram illustrating an exemplary result 5 of speed estimation by an estimation unit according to the third embodiment. FIG. 23 is a diagram illustrating the result of FFT analysis of the speed estimation result illustrated in FIG. 22. 10 FIG. 24 is a diagram illustrating an exemplary configuration of an electric motor system according to a fourth embodiment of the present invention. FIG. 25 is a diagram illustrating an exemplary hardware configuration of a control device according to the 15 fourth embodiment. Description of Embodiments [0011] Hereinafter, an estimation device and an AC motor drive device according to embodiments of the present 20 invention will be described in detail on the basis of the drawings. The present invention is not limited to the embodiments. [0012] First Embodiment. FIG. 1 is a diagram illustrating an exemplary 25 configuration of an electric motor system including an estimation device according to the first embodiment of the present invention. As illustrated in FIG. 1, the electric motor system 100 according to the first embodiment includes an AC motor 1 and a drive device 2 that drives the AC motor 30 1. [0013] The AC motor 1 is a permanent magnet synchronous motor having permanent magnets provided on the rotor, but may be a wound field type synchronous motor having field 9 windings wound around the rotor, or a reluctance type synchronous motor that obtains rotational torque using the saliency of the rotor. In addition, the permanent magnet arrangement of the AC motor 1 may be either embedded type 5 or surface type. In the examples discussed hereinbelow, the AC motor 1 is a three-phase AC motor, but may be an AC motor other than the three-phase AC motor. For example, the AC motor 1 may be a two-phase AC motor or a five-phase AC motor. 10 [0014] The drive device 2 includes a voltage-source inverter 4 and a control device 5. The inverter 4 converts a DC voltage supplied from a DC power supply 3 into an AC voltage and outputs the AC voltage to the AC motor 1. The control device 5 controls the voltage-source inverter 4 to 15 drive the AC motor 1. As illustrated in FIG. 1, the voltage-source inverter 4 includes a main circuit 6 and a gate driver 7. [0015] The main circuit 6 includes a plurality of switching elements Q1, Q2, Q3, Q4, Q5, and Q6. In the main 20 circuit 6, the switching elements Q1, Q3, and Q5 each have one end connected to the high potential side of a DC bus 61, and the switching elements Q2, Q4, and Q6 each have one end connected to the low potential side of the DC bus 61. The other ends of the switching element Q1 and the switching 25 element Q2 are connected to each other to form a U-phase leg. The other ends of the switching element Q3 and the switching element Q4 are connected to each other to form a V-phase leg. The other ends of the switching element Q5 and the switching element Q6 are connected to each other to 30 form a W-phase leg. [0016] The voltage-source inverter 4 includes the threephase bridge circuit including the thus arranged U-phase, V-phase, and W-phase legs. The voltage-source inverter 4 10 can output an AC voltage of a desired amplitude and a desired frequency by switching the plurality of switching elements Q1, Q2, Q3, Q4, Q5, and Q6 between on and off states. The switching elements Q1, Q2, Q3, Q4, Q5, and Q6 5 can be hereinafter collectively referred to as the switching elements Q. [0017] Each of the switching elements Q is an insulated gate bipolar transistor (IGBT) incorporating an antiparallel diode, but may be a metal-oxide-semiconductor 10 field-effect transistor (MOSFET) incorporating an antiparallel diode. In the following examples, the voltagesource inverter 4 is discussed as a two-level inverter that outputs two levels of voltage, but the voltage-source inverter 4 may be a multi-level inverter that outputs three 15 or more levels of voltage. [0018] The gate driver 7 amplifies gate pulse signals Gu, Gv, and Gw output from the control device 5, and outputs the amplified gate pulse signals Gu, Gv, and Gw, which are gate pulse signals Gup, Gvp, and Gwp, to the gates of the 20 switching elements Q1, Q3, and Q5. In addition, the gate driver 7 inversely amplifies the gate pulse signals Gu, Gv, and Gw output from the control device 5, to generate gate pulse signals Gun, Gvn, and Gwn, and outputs the generated Gun, Gvn, and Gwn to the gates of the switching elements Q2, 25 Q4, and Q6. [0019] When one of the gate pulse signals Gup and Gun is in an on state, the other is in an off state. Accordingly, when the switching element Q1 is turned on, the switching element Q2 is turned off, and when the switching element Q1 30 is turned off, the switching element Q2 is turned on. In this manner, the gate pulse signals Gup and Gun cause the switching elements Q1 and Q2 to operate complementarily. Similarly, the gate pulse signals Gvp and Gvn cause the 11 switching elements Q3 and Q4 to operate complementarily, and the gate pulse signals Gwp and Gwn cause the switching elements Q5 and Q6 to operate complementarily. The gate pulse signals Gu, Gv, Gw, Gup, Gun, Gvp, Gvn, Gwp, and Gwn are 5 in the on states when having high potential levels, and are in the off states when having low potential levels. [0020] In addition, the gate driver 7 has a function of insulating the control device 5 which is a low-voltage system from the main circuit 6 which is a high-voltage 10 system, and has a role in preventing failure of the control device 5 in the event of anomaly in the main circuit 6. The gate pulse signals Gu, Gv, and Gw can be hereinafter collectively referred to as the gate pulse signals G. [0021] The control device 5 includes a zero-vector 15 modulation unit 34 that performs zero-vector modulation, a gate pulse generation unit 35 that generates the gate pulse signals Gu, Gv, and Gw, and an estimation device 9 that estimates a magnetic pole position θe and a rotational speed ωe of the rotor of the AC motor 1. The magnetic pole 20 position θe is the electrical angle of the rotor included in the AC motor 1, and is an example of the position of the AC motor 1. The rotational speed ωe is the electrical angular velocity of the rotor included in the AC motor 1, and is an example of the speed of the AC motor 1. The 25 estimation of the magnetic pole position θe may be hereinafter referred to as position estimation, and the estimation of the rotational speed ωe may be hereinafter referred to as speed estimation. [0022] The zero-vector modulation unit 34 irregularly 30 changes the output ratio between two types of zero-voltage vectors that are output from the voltage-source inverter 4. The two types of zero-voltage vectors are a first zerovoltage vector and a second zero-voltage vector. The first 12 zero-voltage vector is output from the voltage-source inverter 4 when all the upper-arm switching elements Q1, Q3, and Q5 are in the on states. The second zero-voltage vector is output from the voltage-source inverter 4 when 5 all the upper-arm switching elements Q1, Q3, and Q5 are in the off states. [0023] By irregularly changing the output ratio between the first zero-voltage vector and the second zero-voltage vector, the spectral peak of carrier noise is dispersed so 10 that carrier noise can be reduced. Carrier noise is a noise that occurs as the AC motor 1, the voltage-source inverter 4, or the like vibrates in accordance with the carrier frequency, i.e. the frequency of a carrier wave Sc to be described later. The output ratio between the first 15 zero-voltage vector and the second zero-voltage vector varies depending on the modulation scheme that the gate pulse generation unit 35 uses. Even when the modulation scheme is changed, the output ratio between the first zerovoltage vector and the second zero-voltage vector can be 20 changed by adding the same values to three-phase voltage commands vu *, vv *, and vw *. The voltage command vu * is a uphase voltage command, the voltage command vv * is a v-phase voltage command, and the voltage command vw * is a w-phase voltage command. 25 [0024] Adding the same positive values to all of the three-phase voltage commands vu *, vv *, and vw * extends the length of time during which all of the gate pulse signals Gu, Gv, and Gw are in the on states, and extends the length of time during which the first zero-voltage vector is 30 output from the voltage-source inverter 4. In addition, adding the same negative value to all of the three-phase voltage commands vu *, vv *, and vw * extends the length of time during which all of the gate pulse signals Gu, Gv, and Gw 13 are in the off states, and extends the length of time during which the second zero-voltage vector is output from the voltage-source inverter 4. [0025] The zero-vector modulation unit 34 performs zero5 vector modulation that irregularly changes the output ratio between the first zero-voltage vector and the second zerovoltage vector by adding a random number value to the three-phase voltage commands vu *, vv *, and vw *. Note that if a preset condition is not satisfied, the zero-vector 10 modulation unit 34 can output the input three-phase voltage commands vu *, vv *, and vw * as they are to the gate pulse generation unit 35. The preset condition is, for example, that the AC motor 1 is rotating at a preset speed or less. In the presence of settings that do not allow zero-vector 15 modulation, the zero-vector modulation unit 34 can output the input three-phase voltage commands vu *, vv *, and vw * as they are to the gate pulse generation unit 35. [0026] The gate pulse generation unit 35 generates the gate pulse signals Gu, Gv, and Gw on the basis of the voltage commands vu *, vv *, and vw * 20 . The gate pulse generation unit 35 outputs the generated gate pulse signals Gu, Gv, and Gw to the gate driver 7 of the voltage-source inverter 4. [0027] The gate pulse generation unit 35 includes a 25 comparison unit 21 and a pulse shift processing unit 22. The comparison unit 21 compares the carrier wave Sc which is a high-frequency periodic signal with the three-phase voltage commands vu *, vv *, and vw *. The pulse shift processing unit 22 performs shift processing on the gate 30 pulse signals Gu, Gv, and Gw output from the comparison unit 21. In the examples discussed herein, the carrier wave Sc is a signal of a triangular wave, but may be a signal of a different waveform such as a sawtooth wave. The carrier 14 wave Sc is also referred to as the carrier signal. [0028] The comparison unit 21 turns on the gate pulse signal Gu when the instantaneous value of the voltage command vu * is less than or equal to the instantaneous 5 value of the carrier wave Sc. The comparison unit 21 turns off the gate pulse signal Gu when the instantaneous value of the voltage command vu * is larger than the instantaneous value of the carrier wave Sc. The comparison unit 21 turns on the gate pulse signal Gv when the instantaneous value of the voltage command vv * 10 is less than or equal to the instantaneous value of the carrier wave Sc. The comparison unit 21 turns off the gate pulse signal Gv when the instantaneous value of the voltage command vv * is larger than the instantaneous value of the carrier wave Sc. The 15 comparison unit 21 turns on the gate pulse signal Gw when the instantaneous value of the voltage command vw * is less than or equal to the instantaneous value of the carrier wave Sc. The comparison unit 21 turns off the gate pulse signal Gw when the instantaneous value of the voltage command vw * 20 is larger than the instantaneous value of the carrier wave Sc. [0029] In the above-described example, the gate pulse generation unit 35 generates the gate pulse signals Gu, Gv, and Gw, using carrier comparison modulation, but the method 25 of generating the gate pulse signals Gu, Gv, and Gw is not limited to that carrier comparison modulation. For example, the gate pulse generation unit 35 may generate the gate pulse signals Gu, Gv, and Gw, using another modulation scheme such as spatial vector modulation, instead of the 30 carrier comparison modulation. In addition, the gate pulse generation unit 35 may generate the gate pulse signals Gup, Gun, Gvp, Gvn, Gwp, and Gwn instead of the gate pulse signals Gu, Gv, and Gw. In this case, the gate driver 7 amplifies 15 the gate pulse signals Gup, Gun, Gvp, Gvn, Gwp, and Gwn. The gate driver 7 outputs the amplified gate pulse signals Gup, Gun, Gvp, Gvn, Gwp, and Gwn to the gates of the switching elements Q1, Q2, Q3, Q4, Q5, and Q6. 5 [0030] The pulse shift processing unit 22 performs the pulse shift processing for shifting the timing at which to switch the gate pulse signal Gu, Gv, or Gw between the on state and the off state. The pulse shift processing by the pulse shift processing unit 22 will be described in detail 10 later. [0031] The estimation device 9 estimates the magnetic pole position θe and the rotational speed ωe of the AC motor 1 on the basis of the gate pulse signals Gu, Gv, and Gw and the voltage commands vu *, vv *, and vw *. The 15 estimation device 9 can also estimate either the magnetic pole position θe or the rotational speed ωe of the AC motor 1. [0032] The estimation device 9 includes a bus current detection unit 11, a phase current determination unit 12, 20 and a time difference calculation unit 13. The bus current detection unit 11 detects the value of a bus current ibus. The phase current determination unit 12 determines the values of phase currents iu, iv, and iw. The time difference calculation unit 13 calculates a time difference 25 Tb (described later) on the basis of the gate pulse signals Gu, Gv, and Gw. In addition, the estimation device 9 includes an estimation unit 15. The estimation unit 15 estimates the magnetic pole position θe and the rotational speed ωe on the basis of the time difference Tb calculated 30 by the time difference calculation unit 13, the values of the phase currents iu, iv, and iw determined by the phase current determination unit 12, and the voltage commands vu *, vv *, and vw *. 16 [0033] The value of the bus current ibus detected by the bus current detection unit 11 is the instantaneous value of the bus current, i.e. the current flowing through the DC bus 61 between the DC power supply 3 and the voltage-source 5 inverter 4. The bus current detection unit 11 may be a current sensor of a type using a current transformer called CT or a current sensor of a type using a shunt resistor. The value of the bus current ibus may be hereinafter simply referred to as the bus current ibus. 10 [0034] In the example illustrated in FIG. 1, the bus current detection unit 11 is provided on the low potential side of the DC bus 61 and detects the value of the current flowing through the low potential side of the DC bus 61, but may be provided on the high potential side of the DC 15 bus 61. In a case where the bus current detection unit 11 is a current sensor of a type using a shunt resistor, providing the bus current detection unit 11 on the low potential side of the DC bus 61 is advantageous in reducing the cost of insulating circuit parts in the bus current 20 detection unit 11. [0035] The phase current determination unit 12 determines the values of the phase currents iu, iv, and iw on the basis of the value of the bus current ibus detected by the bus current detection unit 11. The values of the 25 phase currents iu, iv, and iw determined by the phase current determination unit 12 are the instantaneous values of three phase currents, namely u-phase, v-phase, and wphase currents flowing between the voltage-source inverter 4 and the AC motor 1. Hereinafter, the phase current iu 30 may be referred to as the u-phase current iu, the phase current iv may be referred to as the v-phase current iv, and the phase current iw may be referred to as the w-phase current iw. The values of the phase currents iu, iv, and iw 17 may be simply referred to as the phase currents iu, iv, and iw. [0036] FIGS. 2 to 4 are diagrams for explaining a method for determining the values of phase currents by the phase 5 current determination unit according to the first embodiment. In the examples illustrated in FIGS. 2 to 4, the voltage-source inverter 4 is connected to a Y-connected three-phase resistive load 1a. The current flowing from the point of connection between the switching elements Q1 10 and Q2 to the three-phase resistive load 1a is the u-phase current iu, and the current flowing from the point of connection between the switching elements Q3 and Q4 to the three-phase resistive load 1a is the v-phase current iv. The current flowing from the point of connection between 15 the switching elements Q5 and Q6 to the three-phase resistive load 1a is the w-phase current iw. [0037] The direction of the phase currents flowing from the voltage-source inverter 4 to the three-phase resistive load 1a is the positive direction, and the direction of the 20 phase currents flowing from the three-phase resistive load 1a to the voltage-source inverter 4 is the negative direction. For example, when the u-phase current iu flows in the direction of the arrow illustrated in FIG. 2, the direction of the u-phase current iu is the positive 25 direction. Similarly, when the v-phase current iv flows in the direction of the arrow illustrated in FIG. 2, the direction of the v-phase current iv is the positive direction, and when the w-phase current iw flows in the direction of the arrow illustrated in FIG. 2, the direction 30 of the w-phase current iw is the positive direction. [0038] The phase current determination unit 12 determines the values of the phase currents iu, iv, and iw on the basis of the value of the bus current ibus detected 18 by the bus current detection unit 11 and the states of the plurality of gate pulse signals Gu, Gv, and Gw. For example, the phase current determination unit 12 determines the values of the phase currents iu, iv, and iw on the basis of 5 the value of the bus current ibus detected by the bus current detection unit 11 at the timing of a specific combination pattern of the on and off states of the six switching elements Q. [0039] For example, suppose that the switching elements 10 Q1, Q4, and Q6 are in the on states and the switching elements Q2, Q3, and Q5 are in the off states. In this case, as illustrated in FIG. 3, the u-phase current iu flows from the voltage-source inverter 4 to the three-phase resistive load 1a, and a current having the same magnitude 15 as the u-phase current iu is divided into the v-phase current iv and the w-phase current iw which flow from the three-phase resistive load 1a to the voltage-source inverter 4. The current flowing through the DC bus 61 has the same magnitude as the u-phase current iu; therefore, 20 the phase current determination unit 12 can determine the value of the u-phase current iu from the value of the bus current ibus detected by the bus current detection unit 11 in the state illustrated in FIG. 3. [0040] As illustrated in FIG. 3, the direction of the 25 current flowing through the DC bus 61 is opposite to the direction of the u-phase current iu. The bus current detection unit 11 is located on the DC bus 61 so as to output the positive bus current ibus when the current flowing through the DC bus 61 is in the direction 30 illustrated in FIG. 3. Alternatively, the bus current detection unit 11 may be located on the DC bus 61 so as to output the negative bus current ibus when the current flowing through the DC bus 61 is in the direction 19 illustrated in FIG. 3. In this case, the polarity of the value of the bus current ibus is inverted by the phase current determination unit 12. [0041] Although FIG. 3 depicts an example in which the 5 value of the u-phase current iu in the positive direction is determined, the value of the u-phase current iu in the negative direction, the values of the v-phase current iv in the positive and negative directions, and the values of the w-phase current iw in the positive and negative directions 10 are also determined similarly. For example, suppose that the switching elements Q1, Q3, and Q6 are in the on states and the switching elements Q2, Q4, and Q5 are in the off states as illustrated in FIG. 4. In this case, the phase current determination unit 12 can determine the value of 15 the w-phase current iw in the negative direction from the value of the bus current ibus detected by the bus current detection unit 11. [0042] If the values of two out of the three phase currents iu, iv, and iw are known, the value of the 20 remaining one phase current can be calculated in accordance with Kirchhoff's current law. The phase current determination unit 12 therefore determines the values of any two of the three phase currents iu, iv, and iw from the values of the bus current ibus detected at two different 25 timings by the bus current detection unit 11. Then, the phase current determination unit 12 determines the value of the remaining one phase current through calculation from the determined values of the two phase currents. In this manner, the phase current determination unit 12 determines 30 the values of the three phase currents iu, iv, and iw through a set of two detections of bus currents. [0043] In the case where the voltage-source inverter 4 is a two-level inverter, the six switching elements Q have 20 eight patterns of a combination of on and off states. Six out of these eight combination patterns are each the specific pattern as described above. When the six switching elements Q are in the specific combination 5 pattern of the on and off states, one of the values of the phase currents iu, iv, and iw can be determined from the value of the bus current ibus detected by the bus current detection unit 11. As the remaining two combination patterns allow the voltage-source inverter 4 to output the 10 above-described zero-voltage vectors, it is difficult to determine the values of the phase currents iu, iv, and iw from the value of the bus current ibus detected by the bus current detection unit 11. [0044] FIG. 5 is a diagram illustrating an exemplary 15 relationship between timings of detection of the bus current for use in phase current determination by the phase current determination unit, the carrier wave, the voltage commands, and the gate pulse signals, according to the first embodiment. In FIG. 5, Tc represents a carrier 20 period, i.e. the period of the carrier wave Sc, and fc represents the carrier frequency, i.e. the frequency of the carrier wave Sc. [0045] In the example illustrated in FIG. 5, among the three-phase voltage commands vu *, vv *, and vw *, the voltage command vu * 25 has the largest instantaneous value, the voltage command vv * has the second largest instantaneous value, and the voltage command vw * has the smallest instantaneous value. In the present embodiment, the phase of the voltage command whose absolute value is intermediate among the three-phase voltage commands vu *, vv *, and vw * 30 at a certain point in time is referred to as the intermediate phase. In the example illustrated in FIG. 5, in which the voltage command vv * has an intermediate magnitude, the 21 intermediate phase is the V phase. [0046] In FIG. 5, the length of time from time t1 to time t7 is the first falling half period of the carrier wave Sc. A falling half period is a half period of the 5 carrier wave Sc, during which period the value of the carrier wave Sc gradually decreases. The first falling half period of the carrier wave Sc includes time t3 when the gate pulse signal Gu is in the on state and the gate pulse signals Gv and Gw are in the off states. At time t3, 10 the switching elements Q1, Q4, and Q6 are in the on states and the switching elements Q2, Q3, and Q5 are in the off states. At time t3, therefore, the value of the bus current ibus detected by the bus current detection unit 11 is the same as the value of the u-phase current iu. The 15 phase current determination unit 12 identifies the value of the bus current ibus detected by the bus current detection unit 11 at time t3, as the value of the u-phase current iu. [0047] The first falling half period of the carrier wave Sc includes time t5 when the gate pulse signals Gu and Gv 20 are in the on states and the gate pulse signal Gw is in the off state. At time t5, the switching elements Q1, Q3, and Q6 are in the on states and the switching elements Q2, Q4, and Q5 are in the off states. At time t5, therefore, the value of the bus current ibus detected by the bus current 25 detection unit 11 is the same as the value of the w-phase current iw. The phase current determination unit 12 identifies the value of the bus current ibus detected by the bus current detection unit 11 at time t5, as the value of the w-phase current iw. 30 [0048] The phase current determination unit 12 calculates the value of the v-phase current iv on the basis of Kirchhoff's law from the value of the u-phase current iu obtained from the value of the bus current ibus detected at 22 time t3 and the value of the w-phase current iw obtained from the value of the bus current ibus detected at time t5. The phase current determination unit 12 thus determines the values of the phase currents iu, iv, and iw from the values 5 of the bus current ibus detected by the bus current detection unit 11 at a plurality of current detection timings in the falling half period of the carrier wave Sc. Similarly, the phase current determination unit 12 determines the values of the phase currents iu, iv, and iw 10 using the values of the bus current ibus detected by the bus current detection unit 11 at current detection timings, i.e., times t13 and t15 in the next falling half period from times t11 to t16. [0049] The phase current determination unit 12 thus 15 determines the values of the phase currents iu, iv, and iw from the values of the bus current ibus detected by the bus current detection unit 11 at two current detection timings in a falling half period of the carrier wave Sc. Note that these two current detection timings are not necessarily in 20 a falling half period of the carrier wave Sc. For example, instead of a falling half period of the carrier wave Sc, the phase current determination unit 12 can use a rising half period of the carrier wave Sc to determine the values of the phase currents iu, iv, and iw from the values of the 25 bus current ibus detected by the bus current detection unit 11 at two current detection timings. A rising half period is a half period of the carrier wave Sc, during which period the value of the carrier wave Sc gradually increases. [0050] Immediately after the switching elements Q switch 30 between the on state and the off state, ringing occurs in the bus current ibus due to such switching of the switching elements Q between the on state and the off state. It is difficult to accurately determine the values of the phase 23 currents iu, iv, and iw from the value of the ringing bus current ibus. For this reason, the phase current determination unit 12 waits for a predetermined period of time until the ringing ceases, and thereafter determines 5 the values of the phase currents, using the value of the bus current ibus detected by the bus current detection unit 11. [0051] In the example illustrated in FIG. 5, the phase current determination unit 12 identifies, as the value of 10 the phase current iu, the value of the bus current ibus detected by the bus current detection unit 11 at time t3, which is immediately before time t4 when the gate pulse signal Gv which is the gate pulse signal of the intermediate phase switches from the off state to the on 15 state. In addition, the phase current determination unit 12 identifies, as the value of the phase current iw, the value of the bus current ibus detected by the bus current detection unit 11 at time t5, which is after a preset period of time TA from time t4 when the gate pulse signal 20 Gv switches from the off state to the on state. [0052] The phase current determination unit 12 thus determines a state switching timing at which the gate pulse signal of the intermediate phase among the gate pulse signals Gu, Gv, and Gw switches between the on state and the 25 off state. Then, the phase current determination unit 12 designates each of the timing immediately before the state switching timing and the timing after the period of time TA from the state switching timing, as a bus current detection timing at which the value of the bus current ibus is 30 detected. The phase current determination unit 12 determines the values of any two of the phase currents iu, iv, and iw from the values of the bus current ibus detected by the bus current detection unit 11 at these two bus 24 current detection timings. The phase current determination unit 12 calculates the value of the remaining phase current on the basis of the determined values of the two phase currents. Advantageously, the determination timings of the 5 two phase currents are close to each other, enabling the phase current determination unit 12 to accurately determine the value of the remaining phase current on the basis of the detected values of the two phase currents. [0053] The phase current determination unit 12 can store 10 the values of the bus current ibus repeatedly detected by the bus current detection unit 11. From the stored values of the bus current ibus, the phase current determination unit 12 can extract the value of the bus current ibus detected by the bus current detection unit 11 immediately 15 before the gate pulse signal of the intermediate phase switches. The phase current determination unit 12 can determine the value of the phase current based on the extracted value of the bus current ibus. [0054] Assume that there is a delay time Td from when 20 the gate pulse signal of the intermediate phase switches from the off state to the on state, to when the upper arm of the leg of the intermediate phase switches from the off state to the on state. In this case, the phase current determination unit 12 can also use the value of the bus 25 current ibus detected by the bus current detection unit 11 at the time when the gate pulse signal of the intermediate phase switches between the on state and the off state. In addition to the value of the bus current ibus detected at the time when the gate pulse signal of the intermediate 30 phase switches between the on state and the off state, the phase current determination unit 12 uses the value of the bus current ibus detected a period of time TC after the gate pulse signal of the intermediate phase switches between the 25 on state and the off state. The period of time TC is the period of time TA plus the delay time Td. In the abovedescribed example, the intermediate phase is the v phase. However, the intermediate phase may also be the u phase or 5 the w phase, in which case the phase current determination unit 12 can perform similar processing. [0055] As described above, the phase current determination unit 12 determines two bus current detection timings on the basis of the timing at which the gate pulse 10 signal of the intermediate phase among the plurality of gate pulse signals Gu, Gv, and Gw changes. These two bus current detection timings are timings at which to detect the values of the bus current ibus for use in determining the values of two of the phase currents iu, iv, and iw. The 15 phase current determination unit 12 determines the values of the phase currents iu, iv, and iw on the basis of the plurality of values of the bus current ibus detected by the bus current detection unit 11 every detection time including the determined two bus current detection timings. [0056] Changes in the voltage commands vu *, vv *, and vw * 20 switches the pattern of the combination of the on state and the off state between the above-described eight patterns. The changes in the voltage commands vu *, vv *, and vw * changes the interval at which to switch the combination 25 pattern. The combination pattern is switched by switching of any of the three-phase gate pulse signals Gu, Gv, and Gw between the on state and the off state, immediately after which ringing occurs in the bus current ibus. [0057] The ringing typically converges on the order of 30 several microseconds, but if any of the three-phase gate pulse signals Gu, Gv, and Gw switches between the on state and the off state again before the ringing converges, it is difficult to obtain the pre-switching values of the phase 26 currents. That is, if the state of the gate pulse signal group made up of the three-phase gate pulse signals Gu, Gv, and Gw changes twice in the range of several microseconds, it is difficult to determine the values of the phase 5 currents from the value of the bus current ibus detected by the bus current detection unit 11. Thus, if the interval at which to switch the combination pattern is shorter than the ringing convergence time, it is difficult to determine the values of the phase currents from the value of the bus 10 current ibus detected by the bus current detection unit 11. [0058] In view of this, the pulse shift processing unit 22 of the control device 5 performs a process for shifting the timing at which to switch at least one of the threephase gate pulse signals Gu, Gv, and Gw between the on state 15 and the off state such that the interval at which to switch the combination pattern is longer than or equal to the preset length of time TA. The length of time TA is set to a value longer than or equal to a ringing convergence time that is the length of time from the occurrence of ringing 20 to convergence of ringing. Consequently, the pulse shift processing unit 22 can shift the interval at which to switch between the combination pattern by the ringing convergence time or more, and can secure the latency to the convergence of ringing regardless of the values of the voltage commands vu *, vv *, and vw * 25 . [0059] The pulse shift processing unit 22 performs the pulse shift processing for shifting the timing at which at least one of the three-phase gate pulse signals Gu, Gv, and Gw switches between the on state and the off state within 30 one carrier period. For example, the pulse shift processing unit 22 performs a process for shifting the gate pulse signal G in the length of time made up of the risinghalf and falling-half periods of the carrier wave Sc so 27 that the rising half cycle and the falling half cycle, of the carrier wave Sc have different duty ratios of the gate pulse signal G. [0060] Although the pulse shift processing changes the 5 three-phase voltages output from the voltage-source inverter 4 in a half period of the carrier wave Sc, the duty ratio of the gate pulse signal G in one carrier period is the same before and after shifting the timing of switching between the on state and the off state. 10 Consequently, the switching elements Q1, Q2, Q3, Q4, Q5, and Q6 are turned on and off such that the average of the three-phase voltages output from the voltage-source inverter 4 for each carrier period matches the voltage commands vu *, vv *, and vw *. The pulse shift processing unit 15 22 can therefore prevent a change in the output three-phase voltages of the voltage-source inverter 4 on a carrierperiod-by-carrier-period basis. [0061] The pulse shift processing enables the phase current determination unit 12 to determine the values of 20 the phase currents iu, iv, and iw from the values of the bus current ibus detected by the bus current detection unit 11 regardless of the values of the voltage commands vu *, vv *, and vw *. [0062] In the example illustrated in FIG. 5, the phase 25 current determination unit 12 determines the values of the three phase currents iu, iv, and iw in a falling half period of the carrier wave Sc, but the timings at which to determine the values of the three phase currents iu, iv, and iw are not necessarily in a falling half period of the 30 carrier wave Sc. For example, instead of a falling half period of the carrier wave Sc, the phase current determination unit 12 can use a rising half period of the carrier wave Sc to determine the values of the phase 28 currents iu, iv, and iw on the basis of the values of the bus current ibus detected by the bus current detection unit 11 at two bus current detection timings. [0063] In addition, the phase current determination unit 5 12 can alternately perform the determination process for determining the values of the phase currents iu, iv, and iw in a falling half period of the carrier wave Sc and the determination process for determining the values of the phase currents iu, iv, and iw in a rising half period of the 10 carrier wave Sc. In this case, the phase current determination unit 12 determines the values of the phase currents iu, iv, and iw once every 1.5 times or more a period of the carrier wave Sc because the pulse shift processing unit 22 prevents a change in the output voltages 15 of the voltage-source inverter 4 on a carrier-period-bycarrier-period basis. [0064] In addition, instead of determining the values of the three phase currents iu, iv, and iw in every period of the carrier wave Sc, the phase current determination unit 20 12 can determine the values of the three phase currents iu, iv, and iw in every two or more periods of the carrier wave Sc. In a case where there is an upper limit on the frequency of the carrier wave Sc, the longer three-phase current determination period, which is the period of the 25 determination process for determining the values of the three phase currents iu, iv, and iw results in the frequencies of the phase currents iu, iv, and iw approaching the frequency of the carrier wave Sc during high-speed rotation of the AC motor 1. In this case, the time 30 resolution relative to the one-period waveforms of the phase currents iu, iv, and iw determined by the phase current determination unit 12 is lowered. The lower the time resolution, the lower the estimation accuracy of the 29 estimation device 9 and the control performance by the control device 5. It is therefore preferable that the three-phase current determination period be not long during high-speed rotation. In the presence of the upper limit on 5 the frequency of the carrier wave Sc, therefore, it may be desirable that the three-phase current determination period be one period of the carrier wave Sc. [0065] The above-described state switching timing of the gate pulse signal of the intermediate phase changes in 10 accordance with the voltage command of the intermediate phase. The bus current detection timings at which the values of the bus current ibus are detected change in accordance with the state switching timing of the gate pulse signal of the intermediate phase. The bus current 15 detection timings also change due to the pulse shift processing by the pulse shift processing unit 22. The change of the bus current detection timings due to the pulse shift processing is small, but when the voltage command of the intermediate phase greatly fluctuates, the 20 bus current detection timings also greatly fluctuate. [0066] In the example illustrated in FIG. 5, the value of the v-phase voltage command vv*, which is the voltage command of the intermediate phase, greatly changes at time t11 when the carrier wave Sc reaches the maximum value. As 25 a result, the switching timing of the gate pulse signal Gv, which is the gate pulse signal of the intermediate phase, also greatly varies. Accordingly, there is a large difference between the length of time from time t1 when the carrier wave Sc reaches the maximum value to times t3 and 30 t5, and the length of time from time t11 when the carrier wave Sc reaches the maximum value to times t13 and t15. Times t3 and t5 are bus current detection timings. Times t13 and t15 are bus current detection timings. 30 [0067] As discussed above, the bus current detection timings vary from carrier period to carrier period. In view of this, the estimation unit 15 uses the time difference Tb between current detection times in estimating 5 the magnetic pole position θe and the rotational speed ωe, using the values of the phase currents iu, iv, and iw determined by the phase current determination unit 12. A current detection time is a time identified as the detection time at which the bus current detection unit 11 10 detects the bus current ibus for use in a determination process by the phase current determination unit 12. For example, the current detection time is an average time of the detection time of the bus current ibus used by the phase current determination unit 12 to determine the value of the 15 current of a first phase and the detection time of the bus current ibus used by the phase current determination unit 12 to determine the value of the current of a second phase. In this case, the phase current determination unit 12 calculates the current detection time by adding the 20 detection time of the bus current ibus used to determine the value of the current of the first phase and the detection time of the bus current ibus used to determine the value of the current of the second phase and dividing the resultant value by two. The detection time of the bus current ibus is 25 the time when the bus current ibus is detected by the bus current detection unit 11. [0068] Note that the current detection time may be the detection time at which by the bus current detection unit 11 detects the bus current ibus used to determine the value 30 of the current of the first phase. Alternatively, the current detection time may be the detection time at which the bus current detection unit 11 detects the bus current ibus used to determine the value of the current of the 31 second phase. [0069] For example, in FIG. 5, let tu1 represent time t3 when the bus current ibus is detected for use in the initial determination of the value of the u-phase current iu, and 5 let tu2 represent time t13 when the bus current ibus is detected for use in the next determination of the value of the u-phase current iu. In addition, let tw1 represent time t5 when the bus current ibus is detected for use in the initial determination of the value of the w-phase current 10 iw, and let tw2 represent time t15 when the bus current ibus is detected for the next determination of the value of the w-phase current iw. The period of time TA, which is the latency for ringing convergence, is known. Thus, the time difference between time tu1 and time tw1 is known, and the 15 time difference between time tu2 and time tw2 is also known. However, the time difference between time tu1 and time tu2 or the time difference between time tw1 and time tw2 is not known. In other words, the time difference Tb between current detection times is not known. 20 [0070] In view of this, the time difference calculation unit 13 calculates the time difference Tb between current detection times on the basis of the gate pulse signals Gu, Gv, and Gw. Specifically, the time difference calculation unit 13 determines times tu1, tw1, tu2, and tw2 on the 25 basis of the timings at which the gate pulse signal of the intermediate phase among the gate pulse signals Gu, Gv, and Gw switches between the on state and the off state. On the basis of the determined times tu1, tw1, tu2, and tw2, the time difference calculation unit 13 calculates time tavg1 30 that is the intermediate time between time tu1 and time tw1, and calculates time tavg2 that is the intermediate time between time tu2 and time tw2. The time difference calculation unit 13 calculates the time difference between 32 time tavg1 and time tavg2, as the time difference Tb. The time difference calculation unit 13 thus calculates the time difference Tb between the current detection time identified as the detection time of the bus current ibus 5 used in a determination process by the phase current determination unit 12 and the current detection time identified as the detection time of the bus current ibus used in the next determination process by the phase current determination unit 12. 10 [0071] If the period of time TA, which is the latency for ringing convergence, is constant, each of the time difference between time tu1 and time tu2 and the time difference between time tw1 and time tw2 is the same as the time difference between time tavg1 and time tavg2. In this 15 case, the time difference calculation unit 13 can calculate the time difference between time tu1 and time tu2, as the time difference Tb, or can calculate the time difference between time tw1 and time tw2, as the time difference Tb. FIG. 5 depicts an example in which the time difference 20 between time tw1 and time tw2 is calculated as the time difference Tb. [0072] The fluctuation of the time difference Tb will be described more specifically. FIG. 6 is a diagram illustrating an exemplary relationship between the three25 phase voltage commands and the carrier wave during the rotation of the AC motor at low speed, according to the first embodiment. FIG. 7 is a diagram illustrating an exemplary relationship between the three-phase voltage commands and the carrier wave during the rotation of the AC 30 motor rotates at high speed, according to the first embodiment. In FIGS. 6 and 7, the vertical axis represents modulation rate, and the horizontal axis represents time. The three-phase voltage commands vu *, vv *, and vw * 33 illustrated in FIGS. 6 and 7 are not subjected to zerovector modulation by the zero-vector modulation unit 34. [0073] As illustrated in FIG. 6, when the AC motor 1 rotates at low speed, the frequencies of the three-phase voltage commands vu *, vv *, and vw * 5 are sufficiently low relative to the frequency of the carrier wave Sc. In addition, because the speed electromotive force of the AC motor 1 increases in proportion to the speed of the AC motor 1, when the AC motor 1 rotates at low speed, the amplitudes of the three-phase voltage commands vu *, vv * 10 , and vw * are relatively small. For this reason, the three-phase voltage commands vu *, vv *, and vw * have a gentle slope, and the voltage command of the intermediate phase has a gentle slope. Note that in the example illustrated in FIG. 6, the 15 intermediate phase is the v phase from the first half to a part of the second half, and is the u phase in the remaining part of the second half. [0074] As illustrated in FIG. 7, when the AC motor 1 rotates at high speed, the frequencies of the three-phase voltage commands vu *, vv *, and vw * 20 are higher than when the AC motor 1 rotates at low speed. For example, when the AC motor 1 is rotated by asynchronous pulse width modulation (PWM), it is not uncommon that the ratio of the frequencies of the three-phase voltage commands vu *, vv *, and vw * to the 25 frequency of the carrier wave Sc is about 10% to 15%. When the AC motor 1 rotates at high speed, as illustrated in FIG. 7, the intermediate phase changes at high speed in the order of the v phase, the u phase, the w phase, the v phase,..., and the speed electromotive force is also large, 30 as compared with the case where the AC motor 1 rotates at low speed. For this reason, the amplitudes of the threephase voltage commands vu *, vv *, and vw * are also large as compared with the case illustrated in FIG. 6, and the slope 34 of the voltage command of the intermediate phase is also significantly large. The time difference Tb between current detection times therefore fluctuates greatly. [0075] The higher the time resolution with respect to 5 the waveforms of the phase currents iu, iv, and iw, the better for high-speed rotation of the AC motor 1. In the example illustrated in FIG. 5, the phase current determination unit 12 performs the process of determining the values of the three-phase currents iu, iv, and iw on the 10 basis of the values of the bus current ibus detected by the bus current detection unit 11 in every period of the carrier wave Sc. The fluctuation of the time difference Tb between current detection times relative to the average time difference between current detection times is largest 15 when the values of the three phase currents iu, iv, and iw are determined once every period of the carrier wave Sc. In this case, the time difference Tb between current detection times varies in the range of 0.5×Tbave