Motor control device and air conditioner

Abstract

This motor control device comprises a parameter estimator having: a normative model value calculator that calculates instantaneous reactive power obtained by superimposing a high-frequency command value on a d-axis current command value on the basis of a voltage command value and a motor current, and, in accordance with a prescribed model representing the instantaneous reactive power, calculates a plurality of normative model values for estimating a plurality of motor parameters from the result of calculating the instantaneous reactive power; a mathematical model value calculator that calculates a plurality of mathematical model values corresponding to the plurality of normative model values by using the plurality of motor parameters in accordance with the prescribed model; and an estimation processor that estimates the plurality of motor parameters on the basis of the normative model values and the mathematical model values. The normative model value calculator calculates, as one of the plurality of normative model values, a component including only a single motor parameter in the prescribed model from at least the result of calculating the instantaneous reactive power.

Description

Motor Control Device and Air Conditioner

[0001] The present invention relates to a motor control device that estimates motor parameters and an air conditioner.

[0002] Conventionally, methods for controlling a motor using motor parameters such as motor inductance and armature flux linkage are known. For example, in maximum torque per ampere (MTPA) control, a desired torque can be output with the minimum current amplitude by calculating a motor control command value in accordance with an MTPA curve expressed using motor parameters.

[0003] On the other hand, motor parameters such as inductance and armature flux linkage may vary depending on the operating conditions of the motor. For example, inductance may decrease due to magnetic saturation, and armature flux linkage may fluctuate due to temperature changes. Such changes in motor parameters may cause the MTPA operating point to deviate from the optimal value when, for example, MTPA control is performed, resulting in increased copper loss and degraded operating efficiency. For this reason, methods for correcting the MTPA operating point in accordance with the operating conditions of the motor have been developed.

[0004] For example, Non-Patent Document 1 describes an online method for identifying motor parameters including armature flux linkage, d-axis inductance, and q-axis inductance during motor operation. In this method, a plurality of motor parameters are sequentially estimated using instantaneous reactive power generated by superimposing a high-frequency component on the d-axis current. Specifically, armature flux linkage is estimated by setting the d-axis current to zero, d-axis inductance is then estimated using the estimated value of armature flux linkage, and q-axis inductance is further estimated using the respective estimated values of armature flux linkage and d-axis inductance. By using the motor parameters estimated in this way, it is possible to correct the MTPA operating point onto the optimal MTPA curve, thereby improving torque output characteristics.

[0005] Yuki Kumakiri, et al., "Online Parameter Identification Method for IPM Motors Using Instantaneous Reactive Power", IEEJ Technical Committee Material. MD / Motor Drive Committee [Ed.], November 20, 2014

[0006] On the other hand, the method described in Non-Patent Document 1 requires the sequential estimation of motor parameters such as armature flux linkage, d-axis inductance, and q-axis inductance, which may be time-consuming. Furthermore, since the d-axis current must be set to zero, flux weakening control, which controls the d-axis current to a negative value, becomes impossible. For example, in the MTPA control region near the flux weakening control region (voltage saturation region), setting the d-axis current to zero reduces the field weakening effect, making voltage saturation more likely and potentially making it difficult to maintain proper motor control. For these reasons, the operating region in which the motor parameter estimation process can be performed is limited.

[0007] In view of the above circumstances, the object of the present invention is to provide a motor control device capable of estimating motor parameters in a short time over a wider operating range, and an air conditioner equipped with the same.

[0008] To achieve the above objective, a motor control device according to one embodiment of the present invention is a motor control device that estimates a plurality of motor parameters relating to a motor, and comprises a current command value generator, a high-frequency command value generator, a voltage command value generator, a motor current detector, and a parameter estimator. The current command value generator generates a q-axis current command value and a d-axis current command value based on the rotational speed command value of the motor. The high-frequency command value generator generates a high-frequency command value superimposed on the d-axis current command value. The voltage command value generator generates a voltage command value for driving the motor based on the q-axis current command value and the d-axis current command value on which the high-frequency command value is superimposed. The motor current detector detects the motor current flowing through the motor. The parameter estimator includes a normative model value calculator that calculates instantaneous reactive power by superimposing the high-frequency command value on the d-axis current command value based on the voltage command value and the motor current, and calculates a plurality of normative model values ​​for estimating each of the plurality of motor parameters from the calculation result of the instantaneous reactive power according to a predetermined model representing the instantaneous reactive power; a mathematical model value calculator that calculates a plurality of mathematical model values ​​corresponding to the plurality of normative model values ​​using the plurality of motor parameters according to the predetermined model; and an estimation processor that estimates the plurality of motor parameters based on the plurality of normative model values ​​and the plurality of mathematical model values. The normative model value calculator calculates a component that includes only a single motor parameter in the predetermined model from the calculation result of the instantaneous reactive power as one of the plurality of normative model values.

[0009] In this motor control device, multiple corresponding normative model values ​​and mathematical model values ​​are calculated according to a predetermined model representing instantaneous reactive power generated by superimposing a high-frequency command value on the d-axis current command value. Multiple motor parameters are then estimated using these calculated normative model values ​​and mathematical model values. Furthermore, a component containing only a single motor parameter in the predetermined model is calculated as one of the multiple normative model values. This makes it possible to independently estimate a single motor parameter without restricting, for example, the d-axis current. In addition, it is possible to perform estimation processing for other motor parameters in parallel with the estimation processing of a single motor parameter, while sequentially reflecting the estimation results. This makes it possible to estimate motor parameters over a wider operating range in a shorter time.

[0010] The estimation processor may perform multiple estimation processes in parallel to estimate the multiple motor parameters.

[0011] This makes it possible to estimate multiple motor parameters simultaneously, thereby reducing the time required to estimate each parameter.

[0012] The predetermined model may be a transient model of instantaneous reactive power that includes an in-phase component of the high-frequency command value and an orthogonal component that is generated in accordance with the time change of the current due to the high-frequency command value and is orthogonal to the high-frequency command value.

[0013] This makes it possible to represent instantaneous reactive power more accurately by using a transient model, thereby improving the accuracy of motor parameter estimation.

[0014] The plurality of motor parameters may include armature flux linkage, d-axis inductance, and q-axis inductance. In this case, the transient model may include an orthogonal component to the high-frequency command value as a component that includes only the d-axis inductance. The reference model value calculator may also include an orthogonal component separator that separates the calculation result of the instantaneous reactive power into an orthogonal component and an in-phase component to the high-frequency command value, calculates the orthogonal component as a first reference model value used for estimating the d-axis inductance, and calculates the in-phase component as a second reference model value used for estimating the armature flux linkage, and a low-pass filter that calculates the DC component included in the calculation result of the instantaneous reactive power as a third reference model value used for estimating the q-axis inductance.

[0015] This makes it possible to easily calculate the first reference model value, which is represented solely by the d-axis inductance, by using an orthogonal component separator.

[0016] The estimation processor may estimate the motor parameters by integrally controlling the difference between the corresponding normative model values ​​and the mathematical model values.

[0017] This makes it possible to estimate motor parameters with high accuracy.

[0018] The current command value generator may perform maximum torque / current control for the motor based on the estimated values ​​of the plurality of motor parameters estimated by the parameter estimator.

[0019] This makes it possible to precisely correct the operating point in maximum torque / current control, thereby improving torque output characteristics.

[0020] The motor control device may further include a variable component removal processor that removes the variable component corresponding to the high-frequency command value from the estimated value of the motor's rotational speed that is fed back to the current command value generator.

[0021] Since the fluctuation component caused by the high-frequency command value is removed from the rotational speed feedback value, it becomes possible to accurately calculate the q-axis current command value, d-axis current command value, etc. Furthermore, it becomes possible to set a higher frequency for the high-frequency command value, which improves the signal-to-noise ratio in the motor parameter estimation process.

[0022] The motor control device may further include a high-frequency tracking controller that adjusts the voltage command value so that the motor current follows the high-frequency command value.

[0023] This makes it possible to generate voltage command values ​​that properly reflect high-frequency command values. Furthermore, it becomes possible to set a higher frequency for the high-frequency command values, improving the signal-to-noise ratio (SNR) in the motor parameter estimation process.

[0024] The motor control device may further include an orthogonal component calculator that calculates an orthogonal component that is generated in accordance with the time change of the current due to the high-frequency command value and is orthogonal to the high-frequency command value, and a decoupling controller that performs decoupling control with respect to the voltage command value based on the orthogonal component.

[0025] This enables highly accurate decoupling control using the orthogonal component associated with the time variation of current due to high-frequency command values. Furthermore, it becomes possible to set a higher frequency for the high-frequency command values, improving the signal-to-noise ratio in the motor parameter estimation process.

[0026] An air conditioner according to one embodiment of the present invention is an air conditioner equipped with the motor control device, comprising a motor, a refrigerant circuit, and an inverter circuit. The refrigerant circuit includes a compressor driven by the motor. The inverter circuit drives the motor according to the voltage command value.

[0027] As described above, the present invention makes it possible to estimate motor parameters in a short time over a wider operating range. The effects described herein are not necessarily limited, and any of the effects described in this disclosure may be used.

[0028] This is a schematic diagram showing an example configuration of an air conditioner equipped with a motor control device according to the first embodiment of the present invention. This is a block diagram showing an example configuration of the motor control device according to this embodiment. This is a block diagram showing an example configuration of a parameter estimator. This is a block diagram showing an example configuration of an instantaneous reactive power reference model value calculator. This is a block diagram showing an example configuration of an instantaneous reactive power mathematical model value calculator. This is a block diagram showing an example configuration of an integrator for parameter estimation. This is a table showing the motor parameters and operating conditions used in the simulation. This is a graph showing the identification characteristics of armature flux linkage according to the parameter estimation method of the comparative example. This is a graph showing the identification characteristics of d-axis inductance according to the parameter estimation method of the comparative example. This is a graph showing the identification characteristics of q-axis inductance according to the parameter estimation method of the comparative example. This is a graph showing the identification characteristics of d-axis inductance according to the parameter estimation method of this embodiment. This is a graph showing the identification characteristics of armature flux linkage according to the parameter estimation method of this embodiment. This is a block diagram showing an example configuration of a motor control device according to the second embodiment.

[0029] Embodiments of the present invention will be described below with reference to the drawings.

[0030] <First Embodiment> [Air Conditioner] Figure 1 is a schematic diagram showing an example of the configuration of an air conditioner equipped with a motor control device according to the first embodiment of the present invention. As shown in Figure 1, the air conditioner 100 has an indoor unit 1 and an outdoor unit 2.

[0031] The indoor unit 1 is installed and used in an indoor space within a building. The indoor unit 1 includes an indoor heat exchanger 10, an indoor blower 11, and a controller 12. The outdoor unit 2 is installed and used outdoors of the building where the indoor unit 1 is installed, and is connected to the indoor unit 1 via refrigerant piping through which the refrigerant flows. The outdoor unit 2 includes a motor M, a compressor 20, an outdoor heat exchanger 21, an expansion valve (pressure reducer) 22, a four-way valve (flow path switch) 23, an outdoor blower 24, and a motor control device 30.

[0032] The air conditioner 100 also has a refrigerant circuit 25 that includes a compressor 20 driven by a motor M. Specifically, the outdoor heat exchanger 21, the expansion valve 22, the indoor heat exchanger 10, and the four-way valve 23 connected to the compressor 20 are connected in this order to form the refrigerant circuit 25. The four-way valve 23 switches whether the refrigerant discharged from the compressor 20 flows to the indoor heat exchanger 10 side or to the outdoor heat exchanger 21 side.

[0033] For example, during heating operation, high-temperature, high-pressure refrigerant (gas refrigerant) discharged from the compressor 20 of the outdoor unit 2 flows into the indoor heat exchanger 10 of the indoor unit 1 via the four-way valve 23. The high-pressure refrigerant that has exchanged heat with air in the indoor heat exchanger 10 (condenser) condenses and liquefies. Subsequently, the high-pressure liquid refrigerant is depressurized by passing through the expansion valve 22 of the outdoor unit 2, becoming a low-temperature, low-pressure gas-liquid two-phase refrigerant that flows into the outdoor heat exchanger 21. The refrigerant that has exchanged heat with outside air in the outdoor heat exchanger 21 (evaporator) vaporizes. Subsequently, the low-pressure refrigerant is drawn into the compressor 20 via the four-way valve 23.

[0034] For example, during cooling operation, the high-temperature, high-pressure refrigerant discharged from the compressor 20 of the outdoor unit 2 flows into the outdoor heat exchanger 21 via the four-way valve 23. The high-pressure gaseous refrigerant that has exchanged heat with the outside air in the outdoor heat exchanger 21 (condenser) condenses and liquefies. Subsequently, the high-pressure liquid refrigerant is depressurized by passing through the expansion valve 22 of the outdoor unit 2, becoming a low-temperature, low-pressure gaseous two-phase refrigerant, which flows into the indoor heat exchanger 10 of the indoor unit 1. The refrigerant that has exchanged heat with the air in the indoor heat exchanger 10 (evaporator) evaporates and vaporizes. Subsequently, the low-pressure gaseous refrigerant is drawn into the compressor 20 via the four-way valve 23.

[0035] The indoor fan 11 draws air into the indoor unit 1 and supplies the air, which has undergone heat exchange with the refrigerant in the indoor heat exchanger 10, to the room. The outdoor fan 24 draws air into the outdoor unit 2 and discharges the air, which has undergone heat exchange with the refrigerant in the outdoor heat exchanger 21, to the outside.

[0036] The controller 12 controls the operation of the entire air conditioner 100. Specifically, the controller 12 outputs control signals that control the operation of each part, such as the indoor fan 11, the outdoor fan 24, and the motor control device 30. This enables actions such as switching between operating modes (heating operation, cooling operation, etc.) and controlling the operation of the compressor 20 (rotation control of the motor M) according to the set temperature.

[0037] Motor M is a compressor motor that drives the compressor 20. As motor M, a permanent magnet synchronous motor (PMSM) is used, which uses permanent magnets in the rotor and windings in the stator. Typically, an interior permanent magnet (IPM) motor is used as motor M. Motor M is also provided with three-phase windings (coils) to which three-phase AC voltages output from the inverter circuit 32 (described later) are applied. Hereafter, the phases of the three-phase AC will be referred to as U-phase, V-phase, and W-phase. Note that the specific configuration of motor M is not limited, and any type of permanent magnet motor may be used.

[0038] The motor control device 30 is a device that rotates the motor M based on power supplied from the AC power source 3 and controls the rotational operation of the motor M. Specifically, the motor control device 30 performs vector control of the motor M. In vector control, the current flowing through the windings provided on the stator of the motor M is divided into a current component that generates magnetic flux in the rotor of the motor M (d-axis current) and a current component that generates torque in the rotor (q-axis current), and each current component is controlled independently. In this embodiment, the case in which the motor control device 30 performs vector control of the motor M (compressor motor) that drives the compressor 20 is described, but it is also possible to apply the present invention to the vector control of other motors.

[0039] As shown in Figure 1, the motor control device 30 includes a converter circuit 31, an inverter circuit 32, a calculation circuit 33, a current detector 34, and a DC voltage detector 35.

[0040] The converter circuit 31 is an AC-DC converter that converts the AC voltage output from the AC power supply 3 into a DC voltage, and supplies the DC voltage to the inverter circuit 32. The converter circuit 31 may be, for example, a circuit that outputs a constant DC voltage, or a circuit that can boost or lower the DC voltage (a circuit with a variable output voltage). The specific configuration of the converter circuit 31 is not limited, and any circuit capable of supplying the DC voltage necessary for the operation of the inverter circuit 32 can be used.

[0041] The inverter circuit 32 is a DC-AC converter that converts the DC voltage output from the converter circuit 31 into three-phase AC voltage (three-phase AC voltage of U-phase, V-phase, and W-phase), and supplies the converted three-phase AC voltage to the motor M to drive the motor M. Specifically, the inverter circuit 32 drives the motor M according to the voltage command value generated by the calculation circuit 33. The voltage command value is a command value that specifies the voltage value to be supplied to each of the U-phase, V-phase, and W-phase of the motor M, for example. The inverter circuit 32 applies the voltage indicated by the voltage command value to each phase. Based on this voltage, current flows to each phase of the motor M, and the motor M is driven.

[0042] The voltage applied to the motor M by the inverter circuit 32 is controlled using a pulse width modulation (PWM) signal. The voltage command value is a duty cycle command value that specifies the width (duty cycle) of the PWM signal. The specific configuration of the inverter circuit 32 is not limited; for example, any circuit capable of supplying the three-phase AC voltage necessary for the operation of the motor M can be used.

[0043] The arithmetic circuit 33 is a circuit that performs the calculations necessary for controlling the motor M. The arithmetic circuit 33 is constructed using a computer equipped with a CPU (Central Processing Unit), memory, etc. The arithmetic circuit 33 receives input such as the detected values ​​from the current detector 34 and the DC voltage detector 35, as well as commands and setting values ​​transmitted from the controller 12 installed in the indoor unit 1. Based on these inputs, the voltage command value for performing vector control of the motor M is calculated.

[0044] The current detector 34 detects the current value of the motor current flowing through the motor M. Here, the motor current is a drive current that drives the motor M, and is a current flowing through the windings of the motor M. Specifically, the U-phase current, V-phase current, and W-phase current flowing through the windings of the U-phase, V-phase, and W-phase, respectively, are detected as motor currents. Hereinafter, the current values of the motor current detected by the current detector 34 are described as U-phase current value Iu, V-phase current value Iv, and W-phase current value Iw. The detection results (Iu, Iv, Iw) of the current detector 34 are output to the arithmetic circuit 33.

[0045] The DC voltage detector 35 detects the DC voltage supplied to the inverter circuit 32 (the IPM 45 described later) that drives the motor M. That is, it can also be said that the DC voltage detector 35 detects the output voltage of the converter circuit 31. Hereinafter, the DC voltage supplied to the inverter circuit 32 is described as Vdc. The detection result of the DC voltage detector 35 is output to the arithmetic circuit 33. As the DC voltage detector 35, for example, any voltage sensor suitable for the magnitude of the DC voltage Vdc or the like can be used. [Configuration of Motor Control Device]

[0046] Fig. 2 is a block diagram showing a configuration example of the motor control device according to the present embodiment. Note that illustration of the converter circuit 31 and the DC voltage detector 35 is omitted in Fig. 2.

[0047] The motor control device 30 estimates a plurality of motor parameters related to the motor M, and controls the rotating operation of the motor M using each of the estimated motor parameters. The plurality of motor parameters estimated by the motor control device 30 include armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq.

[0048] The motor control device 30 includes a current command value generator 40, a high-frequency command value generator 41, a voltage command value generator 42, and a d-q / u,v,w converter 43. The motor control device 30 also includes a PWM modulator 44 and an IPM (Intelligent Power Module) 45. Furthermore, the motor control device 30 includes a position sensor 46 (PS), a position / velocity detector 47, a current detection sensor 48, a 3φ current calculator 49, a u,v,w / d-q converter 50, a parameter estimator 51, and a 1 / Pn processor 52. Of these, elements other than the PWM modulator 44, IPM 45, position sensor 46 (PS), and current detection sensor 48 are functional blocks configured, for example, by the calculation circuit 33 described above.

[0049] The current command value generator 40 generates command values ​​for the d-axis current and q-axis current of the motor M so that the motor M rotates at the rotation speed command value (mechanical angular velocity command value ωm*) output from the higher-level controller 12. The current command value generator 40 includes a subtractor 60, a speed controller 61, and an MTPA current command value calculator 62.

[0050] The subtractor 60 calculates the angular velocity deviation (ωm* - ωm) by subtracting the detected mechanical angular velocity value ωm, which is the current mechanical angular velocity of the motor M output by the 1 / Pn processor 52, from the mechanical angular velocity command value ωm*, and outputs the angular velocity deviation (ωm* - ωm) to the speed controller 61.

[0051] The speed controller 61 calculates a q-axis current command value iq* such that the angular velocity deviation (ωm* - ωm) output from the subtractor 60 is small, and outputs the q-axis current command value iq* to the MTPA current command value calculator 62, the subtractor 66, and the decoupling controller 69. The speed controller 61 is composed of, for example, a PI controller, and calculates the q-axis current command value iq* using the following equation (1). In equation (1), kp_s and ki_s are the proportional gain and integral gain in the PI control of the speed controller 61.

[0052] The MTPA current command value calculator 62 calculates the d-axis current command value id_mtpa* on the maximum torque current control (MTPA) curve based on the q-axis current command value iq* output from the speed controller 61 and the estimated values ​​of multiple motor parameters output from the parameter estimator 51, which will be described later, and outputs the d-axis current command value id_mtpa* to the adder 64 of the voltage command value generator 42, which will be described later.

[0053] As will be described later, the parameter estimator 51 calculates estimated values ​​for the armature flux linkage Ψa, the d-axis inductance Ld, and the q-axis inductance Lq. These estimated values ​​will be indicated below with a hat (^). The MTPA current command value calculator 62 calculates the d-axis current command value id_mtpa* from the estimated armature flux linkage Ψa^, the estimated d-axis inductance Ld^, and the estimated q-axis inductance Lq^ using the following equation (2).

[0054] In this manner, the current command value generator 40 generates the q-axis current command value iq* and the d-axis current command value id_mtpa* based on the rotational speed command value (mechanical angular velocity command value ωm*) of the motor M.

[0055] The high-frequency command value generator 41 generates a d-axis current high-frequency command value Δidh that is superimposed on the d-axis current command value id_mtpa*. Specifically, the high-frequency command value generator 41 generates a d-axis current high-frequency command value Δidh with amplitude idh and angular frequency ωh using the following equation (3).

[0056] The high-frequency command value generator 41 outputs the d-axis current high-frequency command value Δidh to the adder 64 of the voltage command value generator 42, which will be described later, and outputs the amplitude idh and angular frequency ωh to the parameter estimator 51. In this embodiment, the d-axis current high-frequency command value Δidh corresponds to the high-frequency command value.

[0057] Here, the angular frequency ωh of the d-axis current high-frequency command value Δidh should be set to be greater than the angular frequency of the motor M (typically the mechanical angular velocity of the motor M) and to be in a separate frequency band so as not to affect the motor control. For example, the angular frequency ωh should preferably be 5 to 10 times or more the angular frequency of the motor M.

[0058] Furthermore, the amplitude idh of the d-axis current high-frequency command value Δidh is determined considering how to minimize the power increase in motor parameter estimation and improve the accuracy of parameter estimation. For example, a larger amplitude idh can be expected to improve estimation accuracy, but it will also increase power. Therefore, the amplitude idh is set to the minimum possible within the range where the required estimation accuracy can be obtained.

[0059] In equation (3), the d-axis current high-frequency command value Δidh is assumed to be a cosine wave, but it is not limited to this, and a sine wave can also be used.

[0060] The voltage command value generator 42 generates voltage command values ​​(q-axis voltage command value Vq* and d-axis voltage command value Vd*) for driving the motor M based on the q-axis current command value iq* and d-axis current command value id_mtpa* output from the current command value generator 40 and the d-axis current high-frequency command value Δidh output from the high-frequency command value generator 41. The voltage command value generator 42 includes an adder 64, a subtractor 65, a subtractor 66, a current controller 67, a Pn processor 68, a decoupling controller 69, an adder 70, and an adder 71.

[0061] The adder 64 calculates a d-axis current command value id* (= id_mtpa* + Δidh) by adding the d-axis current command value Δidh output from the high-frequency command value generator 41 to the d-axis current command value id_mtpa* on the MTPA curve output from the MTPA current command value calculator 62, and outputs the d-axis current command value id* to the subtractor 65 and the decoupling controller 69.

[0062] The subtractor 65 outputs the d-axis current deviation (id* - id) obtained by subtracting the d-axis current detection value id output from the u,v,w / d-q converter 50 from the d-axis current command value id* output from the adder 64 to the current controller 67. The subtractor 66 also outputs the q-axis current deviation (iq* - iq) obtained by subtracting the q-axis current detection value iq output from the u,v,w / d-q converter 50 from the q-axis current command value iq* output from the speed controller 61 to the current controller 67.

[0063] The current controller 67 calculates a command value for the d-axis voltage that reduces the d-axis current deviation (id* - id) output from the subtractor 65, and outputs the command value for the d-axis voltage to the adder 70. The current controller 67 also calculates a command value for the q-axis voltage that reduces the q-axis current deviation (iq* - iq) output from the subtractor 66, and outputs the command value for the q-axis voltage to the adder 71.

[0064] The current controller 67 is configured as, for example, a PI controller, and calculates the d-axis voltage PI command value Vd_pi as the command value of the d-axis voltage using the following equation (4), and calculates the q-axis voltage PI command value Vq_pi as the command value of the q-axis voltage using the following equation (5). In equation (4), kp_d and ki_d are the d-axis proportional gain and d-axis integral gain in the PI control of the current controller 67. Also, in equation (5), kp_q and ki_q are the q-axis proportional gain and q-axis integral gain in the PI control of the current controller 67.

[0065] The Pn processor 68 calculates the electrical angular velocity command value ωe* by multiplying the mechanical angular velocity command value ωm* output from the controller 12 by the number of pole pairs Pn of the motor M, and outputs the electrical angular velocity command value ωe* to the decoupling controller 69.

[0066] The decoupling controller 69 calculates a d-axis decoupling correction value Vd_ff and a q-axis decoupling correction value Vq_ff to suppress interference of induced voltages associated with changes in rotational speed, d-axis current, and q-axis current, respectively, from the electrical angular velocity command value ωe* output from the Pn processor 68, the d-axis current command value id* output from the adder 64, the q-axis current command value iq* output from the speed controller 61, the estimated armature flux linkage value Ψa^, the estimated d-axis inductance value Ld^, and the estimated q-axis inductance value Lq^ output from the parameter estimator 51, and outputs the d-axis decoupling correction value Vd_ff to the adder 70 and the q-axis decoupling correction value Vq_ff to the adder 71.

[0067] The decoherence controller 69 calculates the d-axis decoherence correction value Vd_ff using equation (6) below, and calculates the q-axis decoherence correction value Vq_ff using equation (7) below.

[0068] Here, in calculating the d-axis decoupling correction value Vd_ff and the q-axis decoupling correction value Vq_ff, the electrical angular velocity command value ωe*, the q-axis current command value iq*, and the d-axis current command value id*, which is superimposed with high frequency (d-axis current high frequency command value Δidh), are used. This makes it possible to suppress external noise contained in each detection value and suppress dq interference of high frequency components, compared to, for example, using each detection value for the decoupling correction value. As a result, it is possible to avoid motor control instability and a decrease in high-frequency current tracking ability due to the superposition of external noise on the feedforward compensation values, the d-axis decoupling correction value Vd_ff and the q-axis decoupling correction value Vq_ff.

[0069] The adder 70 calculates a d-axis voltage command value Vd* by adding the d-axis voltage PI command value Vd_pi output from the current controller 67 and the d-axis decoupling correction value Vd_ff output from the decoupling controller 69, and outputs the d-axis voltage command value Vd* to the d-q / u, v, w converter 43 and the parameter estimator 51. The adder 71 also calculates a q-axis voltage command value Vq* by adding the q-axis voltage PI command value Vq_pi output from the current controller 67 and the q-axis decoupling correction value Vq_ff output from the decoupling controller 69, and outputs the q-axis voltage command value Vq* to the d-q / u, v, w converter 43 and the parameter estimator 51.

[0070] The d-axis voltage command value Vd* and the q-axis voltage command value Vq* are the voltage command values ​​actually used to drive the motor M. In this way, the voltage command value generator 42 generates voltage command values ​​(d-axis voltage command value Vd* and q-axis voltage command value Vq*) to drive the motor M based on the q-axis current command value iq* and the d-axis current command value id*, which is superimposed with the d-axis current high-frequency command value Δidh.

[0071] The d-q / u,v,w converter 43 converts the two-phase voltage command values ​​(d-axis voltage command value Vd* and q-axis voltage command value Vq*) output from the voltage command value generator 42 to three-phase voltage command values ​​(U-phase voltage command value Vu*, V-phase voltage command value Vv*, and W-phase voltage command value Vw*) based on the electrical angular phase θdq, which is the current rotor position of the motor M output from the position and speed detector 47 (described later), and outputs them to the PWM modulator 44.

[0072] The PWM modulator 44 generates a six-phase PWM signal from the U-phase voltage command value Vu*, V-phase voltage command value Vv*, and W-phase voltage command value Vw* output from the d-q / u,v,w converter 43, and a PWM carrier signal (not shown), and outputs the six-phase PWM signal to the IPM 45.

[0073] The IPM 45 converts the DC voltage Vdc supplied from the converter circuit 31 shown in Figure 1 into three-phase AC voltages corresponding to the U-phase, V-phase, and W-phase, respectively, based on the six-phase PWM signal output from the PWM modulator 44, and applies each AC voltage to the U-phase, V-phase, and W-phase windings of the motor M.

[0074] In this embodiment, the PWM modulator 44 and the IPM 45 constitute the inverter circuit 32 shown in Figure 1. The U-phase voltage command value Vu*, V-phase voltage command value Vv*, and W-phase voltage command value Vw* input to the PWM modulator 44 are voltage command values ​​obtained by converting the d-axis voltage command value Vd* and the q-axis voltage command value Vq*. Thus, it can be said that the PWM modulator 44 and the IPM 45 (inverter circuit 32) drive the motor M according to the d-axis voltage command value Vd* and the q-axis voltage command value Vq*. The PWM modulator 44 may also be included in the calculation circuit 33.

[0075] The position sensor 46 is a sensor that detects the rotor position of the motor M and outputs a position detection signal indicating the rotor position to the position / speed detector 47. For the position sensor 46, for example, a sensor composed of a Hall element or a photocoupler can be used.

[0076] The position and velocity detector 47 calculates the current rotor position, which is the electrical angular phase θdq, and the detected electrical angular velocity ωe based on the position detection signal of the motor M output from the position sensor 46. The position and velocity detector 47 also outputs the electrical angular phase θdq to the d-q / u,v,w converter 43 and the u,v,w / d-q converter 50, and outputs the detected electrical angular velocity ωe to the parameter estimator 51 and the 1 / Pn processor 52.

[0077] The current detection sensor 48 is a sensor for detecting the current flowing through the U, V, and W phases of the motor M, and can be configured to include a shunt resistor 55, or to include two current sensors 56a and 56b. The shunt resistor 55 is a resistive element for detecting the bus current in a single-shunt system. The current sensors 56a and 56b are current sensors provided on two of the three wirings corresponding to each phase of the motor M (in this case, the wiring for the U and V phases), and are configured using, for example, a CT (Current Transformer).

[0078] The 3φ current calculator 49 calculates the U-phase current value iu, V-phase current value iv, and W-phase current value iw of the motor M from the detection results of the current detection sensor 48. For example, when the bus current is measured using a single shunt method, the current values ​​for the U-phase, V-phase, and W-phase are calculated from the 6-phase PWM switching information output from the PWM modulator 44 and the bus current measured using the shunt resistor 55. Also, when two current sensors 56a and 56b are used, the current values ​​for two phases (U-phase and V-phase) are detected, and the current value for the remaining one phase (W-phase current value) is calculated from the relationship iu + iv + iw = 0. The 3φ current calculator 49 outputs the U-phase current value iu, V-phase current value iv, and W-phase current value iw calculated in this way to the u,v,w / d-q converter 50.

[0079] The u,v,w / d-q converter 50 converts the three-phase motor current (U-phase current value iu, V-phase current value iv, W-phase current value iw) output by the 3φ current calculator 49 into two-phase motor current (d-axis current detection value id and q-axis current detection value iq) based on the electrical angular phase θdq output by the position and velocity detector 47. The u,v,w / d-q converter 50 then outputs the d-axis current detection value id to the parameter estimator 51 and subtractor 65, and outputs the q-axis current detection value iq to the parameter estimator 51 and subtractor 66.

[0080] In this manner, the motor control device 30 detects two-phase motor currents (d-axis current detection value id and q-axis current detection value iq) as the motor current flowing through the motor M using the current detection sensor 48, the 3φ current calculator 49, and the u,v,w / d-q converter 50, and uses them for estimation processing in the parameter estimator 51. In this embodiment, the motor current detector is configured using the current detection sensor 48, the 3φ current calculator 49, and the u,v,w / d-q converter 50.

[0081] The parameter estimator 51 calculates the estimated armature flux linkage Ψa^, the estimated d-axis inductance Ld^, and the estimated q-axis inductance Lq^ from the d-axis voltage command value Vd output from the adder 70, the q-axis voltage command value Vq output from the adder 71, the d-axis current detection value id and the q-axis current detection value iq output from the u,v,w / d-q converter 50, the electrical angular velocity detection value ωe output from the position / velocity detector 47, and the amplitude idh and angular frequency ωh of the d-axis current high-frequency command value Δidh output from the high-frequency command value generator 41. These estimated values ​​are then output to the MTPA current command value calculator 62 and the decoupling controller 69. Details of the processing by the parameter estimator 51 will be described later.

[0082] The 1 / Pn processor 52 calculates the mechanical angular velocity detection value ωm by dividing the electrical angular velocity detection value ωe output from the position and velocity detector 47 by the number of pole pairs Pn of the motor, and outputs the mechanical angular velocity detection value ωm to the subtractor 60.

[0083] [Instantaneous Reactive Power] In this embodiment, motor parameters are estimated using the instantaneous reactive power Q of the motor M. Therefore, we will first explain instantaneous reactive power Q.

[0084] In this embodiment, instantaneous reactive power Q is expressed by absolute transformation using two-phase currents (d-axis current id and q-axis current iq) and two-phase voltages (d-axis voltage vd and q-axis voltage vq) on a rotating coordinate system represented by the d-axis and q-axis. In this case, instantaneous reactive power Q is defined as the cross product of the two-phase current vector (id, iq) and the two-phase voltage vector (vd, vq), as shown in equation (8) below.

[0085] Furthermore, as a method for representing instantaneous reactive power Q, two-phase currents (α-axis current iα, β-axis current iβ) and two-phase voltages (α-axis voltage vα, β-axis voltage vβ) on a stationary coordinate system represented by the α-axis and β-axis can also be used, as described in Non-Patent Document 1. In addition, instantaneous reactive power Q can also be represented by relative transformation.

[0086] Next, we consider expressing the instantaneous reactive power Q in a form that includes multiple motor parameters (Ψa, Ld, Lq) using the voltage equation of the motor M. First, we will explain the case where a steady-state model is used as the model to represent the instantaneous reactive power Q.

[0087] [Steady-State Model] The steady-state model of instantaneous reactive power Q is obtained by applying the voltage equation of an IPM motor (magnetic embedded motor) in a steady state to equation (8) above. The steady-state model does not include components corresponding to the time derivatives of the d-axis current id and the q-axis current iq. The voltage equation of an IPM motor in a steady state is shown in equation (9). In equation (9), Ra is the resistance value (winding resistance) of the motor M.

[0088] Substituting the d-axis voltage vd and q-axis voltage vq, expressed in equation (9), into equation (8), we obtain the instantaneous reactive power Q in a steady-state model including multiple motor parameters (Ψa, Ld, Lq), as shown in equation (10) below.

[0089] Here, we consider superimposing a high-frequency component with amplitude idh and angular frequency ωh onto the d-axis current id. The d-axis current id with the superimposed high-frequency component is shown in equation (11). Here, id0 is the average d-axis current (DC component) excluding the high-frequency components.

[0090] Substituting the d-axis current id, which has the high-frequency component shown in equation (11) superimposed on equation (10), the instantaneous reactive power Q expressed in the steady-state model is given by equation (12) below.

[0091] As shown in equation (12), the first and second terms of the instantaneous reactive power Q expressed in the steady-state model are DC components. The third term is the oscillation component with angular frequency ωh (ωh component), and the fourth term is the oscillation component with twice the angular frequency ωh (2ωh component).

[0092] [Normative Model Values ​​and Mathematical Model Values ​​in the Steady-State Model] Here, normative model values ​​and mathematical model values ​​are introduced in order to estimate each motor parameter (Ψa, Ld, Lq) from the instantaneous reactive power Q. Normative model values ​​are numerical values ​​that serve as a model (norm) for mathematical model values ​​and are calculated from the actual instantaneous reactive power Q. Mathematical model values ​​are numerical values ​​calculated from mathematical model equations that include the motor parameters.

[0093] In the motor parameter estimation process, a normative model value and a mathematical model value are calculated for each motor parameter, and these two values ​​are compared. If the comparison results do not match, it means that the motor parameter values ​​used to calculate the mathematical model values ​​are incorrect, and the motor parameters are corrected so that they match. This makes it possible to estimate the motor parameters.

[0094] Here, the reference model values ​​for armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq are denoted as QΨa, QLd, and QLq, respectively. Furthermore, the mathematical model values ​​for armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq are denoted as QΨa^, QLd^, and QLq^, respectively. Here, the hat symbol (^) appended to the mathematical model values ​​indicates that the value is estimated by a processor such as a microcontroller.

[0095] Next, we will describe a parameter estimation method as a comparative example. This method estimates each parameter based on the instantaneous reactive power Q expressed by the steady-state model shown in equation (12), and is the method described in Non-Patent Document 1.

[0096] For example, by applying a BPF (band-pass filter) that allows the ωh component to pass through to the instantaneous reactive power Q shown in equation (12), only the third term, which does not include the q-axis inductance Lq, can be extracted. Furthermore, by performing control to set id0 = 0, the only motor parameter included in the third term of equation (12) becomes the armature flux linkage Ψa.

[0097] Therefore, when id0 = 0, the component separated by the BPF from the instantaneous reactive power Q shown in equation (12) can be taken as the reference model value QΨa of the armature flux linkage Ψa. In this case, as shown in equation (13), a mathematical model value QΨa^ corresponding to the reference model value QΨa can be set.

[0098] As shown in equation (13), the mathematical model value QΨa^ is expressed using the estimated armature flux linkage value Ψa^. By integral control so that this mathematical model value QΨa^ matches the normative model value QΨa described above, the estimated armature flux linkage value Ψa^ is corrected. By repeating this correction, the estimated armature flux linkage value Ψa^ can be converged to the true value (the correct value of the armature flux linkage Ψa).

[0099] If the estimated armature flux linkage Ψa^ converges and the armature flux linkage Ψa can be identified, the mismatch of the armature flux linkage Ψa in the third term of equation (12) will be eliminated. In this case, if the DC component id0 of the d-axis current is applied, the third term of equation (12) will include the d-axis inductance Ld and the identified armature flux linkage Ψa.

[0100] Using this, when id0 is applied, the component separated by the BPF from the instantaneous reactive power Q shown in equation (12) can be used as the reference model value QLd of the d-axis inductance Ld. In this case, as shown in equation (14), the mathematical model value QLd^ corresponding to the reference model value QLd can be set. Note that the Ψa shown in equation (14) represents the true value (correct armature linkage axial flux) obtained by converging the estimated armature linkage flux Ψa^, and therefore the hat (^) has been removed.

[0101] The estimated d-axis inductance Ld^ is corrected by integral control so that the mathematical model value QLd^ shown in equation (14) matches the normative model value QLd described above. By repeating this correction, the estimated d-axis inductance Ld^ can be converged to the true value (the correct value of the d-axis inductance Ld).

[0102] If the estimated d-axis inductance Ld^ converges and the d-axis inductance Ld can be identified, the mismatch between the armature flux linkage Ψa and the d-axis inductance Ld is eliminated. In other words, both Ψa and Ld converge to their true values. In this case, the first term, which is the DC component of equation (12), will include the q-axis inductance Lq and the identified armature flux linkage Ψa and d-axis inductance Ld. Such a DC component can be extracted, for example, using an LPF (low-pass filter).

[0103] Therefore, the DC component separated by the LPF from the instantaneous reactive power Q shown in equation (12) can be taken as the reference model value QLq for the q-axis inductance Lq. In this case, as shown in equation (15), the mathematical model value QLq^ corresponding to the reference model value QLq can be set. Although the second term is also extracted as a DC component, this is not considered in Non-Patent Literature 1. In addition, Ld and Ψa shown in equation (15) represent the true values ​​obtained by converging the estimated d-axis inductance Ld^ and the estimated armature flux linkage Ψa^, respectively.

[0104] The estimated q-axis inductance Lq^ is corrected by integral control so that the mathematical model value QLq^ shown in equation (15) matches the normative model value QLq described above. By repeating this correction, the estimated q-axis inductance Lq^ can be converged to the true value (the correct value of the q-axis inductance Lq).

[0105] In the parameter estimation method of the comparative example, the motor parameters, armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq, are identified in this order. In other words, it is necessary to perform the process of identifying each motor parameter sequentially, which has the problem of taking a long time to identify all motor parameters.

[0106] Furthermore, in order to identify the armature flux linkage Ψa, it is necessary to perform control such as setting id0 = 0. id0 refers to the d-axis current id (d-axis current command value id_mtpa* in Figure 2), which is calculated together with the q-axis current iq in MTPA control, for example. However, setting id0 = 0 makes it difficult to perform proper MTPA control.

[0107] For example, in the MTPA control region near the flux weakening control region, voltage saturation can be avoided by performing flux weakening control by setting the d-axis current id (id0 in this case) to a negative value. However, if id0 = 0, flux weakening control becomes impossible, and depending on the operating conditions, voltage saturation may occur, making it difficult to properly control the motor M. For this reason, the parameter estimation method in the comparative example, which requires id0 = 0, cannot be applied to the MTPA control region near the flux weakening control region.

[0108] Furthermore, when identifying the d-axis inductance Ld, it is necessary to increase the average d-axis current id0 to improve the signal-to-noise ratio. Thus, the parameter estimation method of the comparative example presents many practical inconveniences when implemented during MTPA control.

[0109] Furthermore, in the parameter estimation method of the comparative example, the steady-state model shown in equation (9) (voltage equation of motor M in a steady state) is used to construct the model of instantaneous reactive power Q shown in equation (12). In reality, since the d-axis current id is superimposed with the d-axis current high frequency (the second term of equation (11)), a p-term voltage caused by the high frequency should be generated. The p-term voltage is the induced voltage (pLdid) at the d-axis inductance Ld caused by the fluctuation of the d-axis current id. In the notation pLdid, "p" is the time derivative operator (d / dt) and represents the time derivative of the current value "id". Therefore, pLdid = Ld・did / dt.

[0110] The p-term voltage is generated as a voltage component that fluctuates with a phase leading π / 2 relative to the d-axis current high frequency with angular frequency ωh, while the instantaneous reactive power Q is assumed to be generated as an ωh component that fluctuates with a phase lagging π / 2 relative to the d-axis current high frequency. As a result, the ωh component extracted by the BPF is interfered with by the p-term voltage induced by the d-axis inductance Ld. Therefore, even when controlling id=0, the reference model value QΨa extracted from the instantaneous reactive power Q by the BPF will also include the p-term voltage component.

[0111] If the normative model value QΨa, which includes the p-term voltage component, is used as the norm for the mathematical model value QΨa^, which includes only the armature flux linkage Ψa shown in equation (13), then the identified value of the armature flux linkage Ψa may deviate from the true value. Furthermore, the deviation of the armature flux linkage Ψa from the true value makes it difficult for the d-axis inductance Ld and the q-axis inductance Lq to converge to the correct values. As a result, each motor parameter will deviate from its true value, and for example, the MTPA operating point may not be optimal, making it impossible to minimize copper loss in the motor M. Thus, when using a steady-state model, the estimation accuracy of motor parameters may decrease.

[0112] [Transient Model] In order to solve the problem of estimation accuracy when using the steady-state model described above, the inventors investigated transient models. Here, we will explain the case in which a transient model is used as a model to represent instantaneous reactive power Q.

[0113] The transient model of instantaneous reactive power Q is obtained by applying the voltage equation of the IPM motor in the transient state to equation (8) above. The transient model includes components corresponding to the time derivatives of the d-axis current id and the q-axis current iq. The voltage equation of the IPM motor in the transient state is shown in equation (16). The "p" in equation (16) is the differential operator described above.

[0114] Furthermore, the d-axis current id is superimposed with the high-frequency d-axis current Δidh, similar to equation (11). The high-frequency d-axis current Δidh and the d-axis current id are shown in equations (17) and (18).

[0115] Furthermore, the q-axis current iq is assumed to be DC. That is, the q-axis current iq does not contain frequency components that change with time. In this case, the p-term voltages pLdid and pLqiq included in equation (16) are expressed by equations (19) and (20) shown below, respectively.

[0116] Here, substituting the d-axis voltage vd and q-axis voltage vq, expressed in equation (16), into the definition of instantaneous reactive power Q shown in equation (8), we obtain a transient model of instantaneous reactive power Q. Furthermore, using equations (17) to (20), the transient model of instantaneous reactive power Q is given by equation (21) below.

[0117] As shown in equation (21), the transient model of instantaneous reactive power Q is obtained by adding a fifth term (a sin component oscillating at angular frequency ωh), which is the instantaneous reactive power generated by pLdid, to the steady-state model of instantaneous reactive power Q shown in equation (12). This fifth term component is orthogonal to the third term (a cos component oscillating at angular frequency ωh).

[0118] [Normative Model Values ​​and Mathematical Model Values ​​in Transient Models] Next, we will explain how to set normative model values ​​and mathematical model values ​​for the instantaneous reactive power Q shown in equation (21).

[0119] Various methods can be considered for setting the normative model value and mathematical model value for instantaneous reactive power Q. For example, the setting can be done in the same way as the parameter estimation method in the comparative example described above. In this case, we focus on the third term in equation (21), and first set id0 = 0 to disable the component corresponding to the d-axis inductance Ld, thereby focusing only on the armature flux linkage Ψa.

[0120] Thus, in the parameter estimation method of the comparative example, since the third term of interest includes two motor parameters (Ψa and Ld), it is necessary to first focus on estimating the value of one motor parameter (Ψa in this case). At this time, the other motor parameter (Ld in this case) has its corresponding component disabled, so it is not possible to estimate both motor parameters simultaneously. As a result, it is necessary to estimate Ψa, then Ld, and then Lq. Furthermore, the process of disabling the component corresponding to the motor parameter other than the one being estimated (Ld in this case) in the third term (control that sets id0=0) causes a narrowing of the operating range in which the parameter estimation method of the comparative example can be applied.

[0121] Based on these considerations, the inventor focused on a component within the instantaneous reactive power Q that contains only a single motor parameter (hereinafter referred to as the single-parameter component). In this case, each motor parameter is estimated by setting normative model values ​​and mathematical model values ​​based on at least the single-parameter component for the instantaneous reactive power Q.

[0122] For example, by setting normative model values ​​and mathematical model values ​​for a single-parameter component, the motor parameters included in that single-parameter component can be estimated independently. Furthermore, by utilizing the estimated values ​​obtained during the estimation process of the single-parameter component's motor parameters, it becomes possible to simultaneously estimate other motor parameters. This reduces the time required to estimate multiple motor parameters.

[0123] Furthermore, when estimating motor parameters included in a single parameter component, no processing is required to disable components corresponding to other motor parameters. Therefore, it becomes possible to estimate each motor parameter over a wide operating range without performing control such as setting id0=0.

[0124] The principle of the parameter estimation method according to this embodiment will be explained below. Equations (22), (23), and (24) represent the mathematical models QLd for estimating the d-axis inductance Ld, QΨa for estimating the armature flux linkage Ψa, and QLq for estimating the q-axis inductance Lq, respectively. Note that each mathematical model described here is an extraction of the corresponding term from equation (21) for mathematical explanation purposes. Similarly, the normative model values ​​that serve as the basis for each mathematical model value will also be described using the same sign.

[0125] First, let's explain the single-parameter component. As shown in equation (21), the fifth term of the instantaneous reactive power Q expressed by the transient model includes only the d-axis inductance Ld as a motor parameter. Therefore, the fifth term is a single-parameter component. In this embodiment, the reference model value and mathematical model value for estimating the d-axis inductance Ld are set from the fifth term of equation (21).

[0126] For example, when orthogonal separation is performed on the angular frequency ωh component of the instantaneous reactive power Q shown in equation (21), the components of the third and fifth terms are separated. Of these, the component of the fifth term can be taken as the reference model value QLd of the d-axis inductance Ld. In this case, as shown in equation (22), the fifth term of equation (21) can be taken as the mathematical model value corresponding to the reference model value QLd.

[0127] Furthermore, as shown in equation (21), the third term includes the d-axis inductance Ld and the armature flux linkage Ψa. However, since the d-axis inductance Ld can be estimated using the fifth term, the third term essentially becomes a term with a value corresponding to the armature flux linkage Ψa. Using this, in this embodiment, a normative model value and a mathematical model value for estimating the armature flux linkage Ψa are set from the third term of equation (21).

[0128] As described above, when orthogonal separation is performed on the instantaneous reactive power Q shown in equation (21), the third term is separated along with the fifth term. The component of the third term thus separated can be taken as the normative model value QΨa of the armature flux linkage Ψa. In this case, as shown in equation (23), the third term of equation (21) can be taken as the mathematical model value corresponding to the normative model value QΨa.

[0129] Furthermore, in equation (21), the DC component including the first and second terms can be used as the component including the q-axis inductance Lq. Note that the DC component also includes the d-axis inductance Ld and the armature flux linkage Ψa, but since these parameters can be estimated using equations (22) and (23), the first and second terms essentially become terms with values ​​corresponding to the q-axis inductance Lq. Using this, in this embodiment, a normative model value and a mathematical model value for estimating the q-axis inductance Lq are set from the first and second terms of equation (21).

[0130] For example, the DC component can be separated by passing the instantaneous reactive power Q shown in equation (21) through an LPF. The DC components separated in this way (terms 1 and 2) can be taken as the reference model value QLq of the q-axis inductance Lq. In this case, as shown in equation (24), terms 1 and 2 of equation (21) can be taken as mathematical model values ​​corresponding to the reference model value QLq.

[0131] Thus, in the parameter estimation method according to this embodiment, the third and fifth terms are separated by orthogonally separating the angular frequency ωh component from the instantaneous reactive power Q shown in equation (21). Of these, the d-axis inductance Ld is estimated using the fifth term. The armature flux linkage Ψa is estimated using the third term, as in the comparative example, but in this case there is no need to control id=0, and as the d-axis inductance Ld converges to its true value, the armature flux linkage Ψa also converges to its true value. Furthermore, the q-axis inductance Lq is estimated using the first and second terms, and as the d-axis inductance Ld and the armature flux linkage Ψa converge to their true values, the q-axis inductance Lq also converges to its true value. This makes it possible, for example, to simultaneously identify multiple motor parameters (Ψa, Ld, Lq).

[0132] Furthermore, the instantaneous reactive power Q shown in equation (21) is a transient model that takes into account the p-term voltage (pLd) generated by superimposing the high-frequency d-axis current Δidh. This makes it possible to estimate each motor parameter (Ψa, Ld, Lq) with high accuracy.

[0133] [Parameter Estimator] Figure 3 is a block diagram showing an example configuration of a parameter estimator. The parameter estimator 51 includes an instantaneous reactive power reference model value calculator 74, an instantaneous reactive power mathematical model value calculator 75, a subtractor 76, a subtractor 77, a subtractor 78, and a parameter estimation integrator 79.

[0134] The instantaneous reactive power reference model value calculator 74 calculates the instantaneous reactive power Q by superimposing the high-frequency command value Δidh on the d-axis current command value id_mtpa* based on the voltage command value (d-axis voltage command value Vd* and q-axis voltage command value Vq*) and the motor current (d-axis current detection value id and q-axis current detection value iq).

[0135] Furthermore, the instantaneous reactive power reference model value calculator 74 calculates multiple reference model values ​​(QLd, QΨa, QLq) for estimating each of multiple motor parameters (Ld, Ψa, Lq) from the calculation result of instantaneous reactive power Q, according to a predetermined model representing instantaneous reactive power Q.

[0136] In the following, the reference model values ​​QLd, QΨa, and QLq may be referred to as the reference model value QLd for d-axis inductance estimation, the reference model value QΨa for armature flux linkage estimation, and the reference model value QLq for q-axis inductance estimation. In this embodiment, the instantaneous reactive power reference model value calculator 74 corresponds to the reference model value calculator.

[0137] The instantaneous reactive power mathematical model value calculator 75 calculates multiple mathematical model values ​​(QLd^, QΨa^, QLq^) corresponding to multiple normative model values ​​(QLd, QΨa, QLq) using multiple motor parameters (Ld, Ψa, Lq) according to a predetermined model representing instantaneous reactive power Q. The motor parameters (Ld, Ψa, Lq) used in calculating each mathematical model value are estimated values ​​(Ld^, Ψa^, Lq^) fed back from the parameter estimation integrator 79.

[0138] In the following, the mathematical model values ​​QLd^, QΨa^, and QLq^ may be referred to as the mathematical model value QLd^ for d-axis inductance estimation, the mathematical model value QΨa^ for armature flux linkage estimation, and the mathematical model value QLq^ for q-axis inductance estimation. In this embodiment, the instantaneous reactive power mathematical model value calculator 75 corresponds to the mathematical model value calculator.

[0139] Subtractor 76 subtracts the mathematical model value QLd^ for d-axis inductance estimation from the standard model value QLd for d-axis inductance estimation, and outputs the subtraction result (model deviation of d-axis inductance Ld) to the parameter estimation integrator 79. Subtractor 77 subtracts the mathematical model value QΨa^ for armature flux linkage estimation from the standard model value QΨa for armature flux linkage estimation, and outputs the subtraction result (model deviation of armature flux linkage Ψa) to the parameter estimation integrator 79. Subtractor 78 subtracts the mathematical model value QLq^ for q-axis inductance estimation from the standard model value QLq for q-axis inductance estimation, and outputs the subtraction result (model deviation of q-axis inductance Lq) to the parameter estimation integrator 79.

[0140] The parameter estimation integrator 79 performs integration on the model deviations of armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq, calculates the estimated armature flux linkage Ψa^, the estimated d-axis inductance Ld^, and the estimated q-axis inductance Lq^, and feeds each estimated value back to the instantaneous reactive power mathematical model value calculator 75.

[0141] In this embodiment, an estimation processor is realized that estimates multiple motor parameters based on multiple normative model values ​​and multiple mathematical model values, using three subtractors 76, subtractors 77, subtractors 78, and a parameter estimation integrator 79.

[0142] The parameter estimator 51 performs an estimation method that converges the motor parameters to their true values ​​by repeatedly feeding back the estimated motor parameters derived from the normative model values ​​and mathematical model values ​​in this manner.

[0143] As described above, the parameter estimation method according to this embodiment utilizes a single-parameter component as a reference model value and a mathematical model value set from a predetermined model representing instantaneous reactive power Q. Therefore, the instantaneous reactive power reference model value calculator 74 calculates a component (single-parameter component) that includes only a single motor parameter in the predetermined model from the calculation result of instantaneous reactive power Q as one of a plurality of reference model values. In addition, the instantaneous reactive power mathematical model value calculator 75 calculates a mathematical model value corresponding to at least the above single-parameter component as one of a plurality of reference model values. This makes it possible to perform independent estimation processing for at least one parameter among a plurality of motor parameters.

[0144] Furthermore, in this embodiment, the transient model shown in equation (21) is used as a predetermined model representing the instantaneous reactive power Q. The transient model includes an in-phase component of the high-frequency command value Δidh and an orthogonal component that is generated with the time change of the current due to the high-frequency command value Δidh and is orthogonal to the high-frequency command value Δidh. Here, as shown in equation (3), the high-frequency command value Δidh is the cos component of the angular frequency ωh. Therefore, in equation (21), the third term is the in-phase component of the high-frequency command value Δidh, and the fifth term is the orthogonal component. By using the transient model, the estimation accuracy of each motor parameter can be improved.

[0145] [Instantaneous Reactive Power Reference Model Value Calculator] Figure 4 is a block diagram showing an example configuration of an instantaneous reactive power reference model value calculator. The instantaneous reactive power reference model value calculator 74 includes an instantaneous reactive power calculator 80, an orthogonal component separator 81, and an LPF processor 82.

[0146] The instantaneous reactive power calculator 80 calculates the instantaneous reactive power Q based on the two-phase voltages, the d-axis voltage command value Vd* and the q-axis voltage command value Vq*, and the two-phase currents, the d-axis current detection value id and the q-axis current detection value iq, and outputs the calculation result to the orthogonal component separator 81 and the LPF processor 82. Specifically, the instantaneous reactive power Q is calculated according to equation (25). This process calculates the instantaneous reactive power Q according to the definition explained in equation (8).

[0147] The instantaneous reactive power Q shown in equation (25) is a measured value that represents the actual value of the instantaneous reactive power Q generated by the motor M. In contrast, the transient model of the instantaneous reactive power Q shown in equation (21) is a theoretical formula that takes into account the p-term voltage.

[0148] The orthogonal component separator 81 uses the angular frequency ωh output from the high-frequency command value generator 41 to calculate the reference model value QLd for estimating the d-axis inductance and the reference model value QΨa for estimating the armature flux linkage, based on the instantaneous reactive power Q output from the instantaneous reactive power calculator 80.

[0149] As described above, the transient model shown in equation (21) includes an orthogonal component (term 5) with respect to the high-frequency command value Δidh as a component that includes only the d-axis inductance Ld. The transient model also includes an in-phase component (term 3) of the high-frequency command value Δidh as a component that includes the d-axis inductance Ld and the armature flux linkage Ψa. In this embodiment, an orthogonal component separator 81 is provided to calculate these orthogonal and in-phase components.

[0150] The orthogonal component separator 81 separates the calculation result of instantaneous reactive power Q by the instantaneous reactive power calculator 80 into orthogonal and in-phase components with respect to the high-frequency command value Δidh. The orthogonal component separator 81 also calculates the orthogonal component as the d-axis inductance estimation reference model value QLd, which is used to estimate the d-axis inductance Ld, and calculates the in-phase component as the armature linkage flux estimation reference model value QΨa, which is used to estimate the armature linkage flux Ψa. In this embodiment, the d-axis inductance estimation reference model value QLd corresponds to the first reference model value, and the armature linkage flux estimation reference model value QΨa corresponds to the second reference model value.

[0151] Specifically, the orthogonal component separator 81 extracts the sine component (reference model value QLd for d-axis inductance estimation) at an angular frequency ωh orthogonal to the high-frequency command value Δidh from the instantaneous reactive power Q according to equation (26). The orthogonal component separator 81 also extracts the cosine component (reference model value QΨa for armature flux linkage estimation) at an angular frequency ωh that is in phase with the high-frequency command value Δidh from the instantaneous reactive power Q according to equation (27). In equations (26) and (27), "Th" is the period of the high-frequency command value (2π / ωh), and "t" is the instantaneous time.

[0152] In equation (26), the amplitude of the sine component at angular frequency ωh of the instantaneous reactive power Q, which includes high frequencies, is calculated. Similarly, in equation (27), the amplitude of the cosine component at angular frequency ωh of the instantaneous reactive power Q, which includes high frequencies, is calculated. This can also be described as the process of extracting the amplitude components from equations (22) and (23).

[0153] For example, in the parameter estimation method of the comparative example, the effective value (RMS) of the high-frequency component (a component obtained by combining the sin component and the cos component) extracted by the BPF is used as a reference model value for estimating the armature flux linkage Ψa. In contrast, in this embodiment, the amplitude of the high-frequency sin component becomes the reference model value QLd for estimating the d-axis inductance, and the amplitude of the cos component becomes the reference model value QΨa for estimating the armature flux linkage.

[0154] The LPF processor 82 calculates the DC component included in the calculation result of instantaneous reactive power Q by the instantaneous reactive power calculator 80 as a reference model value QLq for estimating the q-axis inductance, which is used to estimate the q-axis inductance Lq. In this embodiment, the reference model value QLq for estimating the q-axis inductance corresponds to the third reference model value, and the LPF processor 82 corresponds to a low-pass filter processor.

[0155] The LPF processor 82 is configured as an LPF that allows the DC component to pass through. Here, assuming a first-order LPF, for example, the reference model value QLq for estimating the q-axis inductance is calculated according to equation (28). In equation (28), "K" is the filter gain in the passband, "T" is the filter time constant, and "s" is a variable of the Laplace transform. The configuration of the LPF processor 82 is not limited, and any filter capable of extracting the DC component can be used.

[0156] [Instantaneous Reactive Power Mathematical Model Value Calculator] Figure 5 is a block diagram showing an example configuration of the instantaneous reactive power mathematical model value calculator. The estimated values ​​of each motor parameter (Ld^, Ψa^, Lq^) input in Figure 5 are the outputs of the parameter estimation integrator 79, as explained with reference to Figure 3. Note that the initial values ​​of Ld^, Ψa^, and Lq^ immediately before the estimation process are set to, for example, Typ values ​​(reference values ​​described in the motor M specifications).

[0157] The instantaneous reactive power mathematical model value calculator 75 calculates the mathematical model value QΨa^ for estimating armature flux linkage, the mathematical model value QLd^ for estimating d-axis inductance, and the mathematical model value QLq^ for estimating q-axis inductance, respectively, based on a transient model of instantaneous reactive power Q including the p-term voltage. As shown in Figure 5, the instantaneous reactive power mathematical model value calculator 75 includes an LPF processor 84, a mathematical model value calculator 85 for estimating d-axis inductance, a mathematical model value calculator 86 for estimating armature flux linkage, and a mathematical model value calculator 87 for estimating q-axis inductance.

[0158] The LPF processor 84 extracts the DC component of the detected d-axis current id and outputs it as the average d-axis current id0. Any filter capable of extracting the DC component can be used as the LPF processor 84.

[0159] The mathematical model value calculator 85 for estimating the d-axis inductance calculates the mathematical model value QLd^, which is the orthogonal component (sin component) of the instantaneous reactive power Q with the d-axis current high-frequency command value Δidh, based on the q-axis current detection value iq, the amplitude idh and angular frequency ωh of the d-axis current high-frequency command value Δidh, and the estimated d-axis inductance value Ld^, according to equation (29). Equation (29) is the formula for calculating the amplitude of the sine component shown in equation (22).

[0160] The mathematical model value calculator 86 for estimating armature flux linkage calculates the mathematical model value QΨa^, which is the in-phase component (cos component) of the d-axis current high-frequency command value Δidh included in the instantaneous reactive power Q, based on the detected electrical angular velocity value ωe, the average d-axis current id0, the amplitude idh of the d-axis current high-frequency command value Δidh, the estimated d-axis inductance value Ld^, and the estimated armature flux linkage value Ψa^, according to equation (30). Equation (30) is the formula for calculating the amplitude of the cos component shown in equation (23).

[0161] The mathematical model value calculator 87 for estimating the q-axis inductance calculates the mathematical model value QLq^, which is the DC component of the instantaneous reactive power Q, according to equation (31), based on the detected electrical angular velocity ωe, the average d-axis current id0, the detected q-axis current iq, the amplitude idh of the d-axis current high-frequency command value Δidh, the estimated d-axis inductance Ld^, the estimated armature flux linkage Ψa^, and the estimated q-axis inductance Lq^. Equation (31) is equivalent to equation (24).

[0162] [Integrator for Parameter Estimation] Figure 6 is a block diagram showing an example configuration of the integrator for parameter estimation. The integrator for parameter estimation 79 estimates motor parameters by integrally controlling the difference between the corresponding normative model values ​​and mathematical model values. The difference between the normative model values ​​and mathematical model values ​​is calculated by the three subtractors 76, 77, and 78 shown in Figure 3. As shown in Figure 6, the integrator for parameter estimation 79 includes a d-axis inductance estimator 88, an armature flux linkage estimator 89, and a q-axis inductance estimator 90.

[0163] As explained below, the processes performed by the d-axis inductance estimator 88, the armature flux linkage estimator 89, and the q-axis inductance estimator 90 are estimation processes that estimate the corresponding motor parameters (d-axis inductance Ld, armature flux linkage Ψa, and q-axis inductance Lq). In the parameter estimation integrator 79, multiple estimation processes for estimating these multiple motor parameters are executed in parallel. This eliminates the need to estimate each motor parameter sequentially, making it possible to estimate multiple motor parameters in a short amount of time.

[0164] The d-axis inductance estimator 88 calculates the estimated d-axis inductance Ld^ by integrally controlling the difference (QLd - QLd^) between the reference model value QLd for d-axis inductance estimation output from the subtractor 76 and the mathematical model value QLd^ for d-axis inductance estimation, using the q-axis current detection value iq and the amplitude idh and angular frequency ωh of the d-axis current high-frequency command value Δidh. Specifically, the estimated d-axis inductance Ld^ is estimated according to equation (32). In equation (32), "G" is the integral gain.

[0165] The mathematical model value QLd^ for estimating the d-axis inductance includes only the d-axis inductance Ld among the motor parameters, as shown in equation (29). This allows the d-axis inductance estimate Ld^ to be estimated independently.

[0166] Specifically, the estimated d-axis inductance Ld^ is adjusted, and the integral operation using equation (32) is repeated until the difference between the normative model value and the mathematical model value (QLd - QLd^) becomes zero. In this process, the amount of integration for that iteration becomes zero when the difference becomes zero, but the result of the integration up to that point is the d-axis inductance Ld. In this way, by modifying the estimated d-axis inductance Ld^ in equation (32) until QLd - QLd^ becomes zero, the estimated d-axis inductance Ld^ can be converged to the true value.

[0167] The armature flux linkage estimator 89 calculates the estimated armature flux linkage Ψa^ by integrally controlling the difference (QΨa - QΨa^) between the reference model value QΨa for armature flux linkage estimation output from the subtractor 77 and the mathematical model value QΨa^ for armature flux linkage estimation, using the amplitude idh of the d-axis current high-frequency command value Δidh and the detected electrical angular velocity value ωe. Specifically, the estimated armature flux linkage Ψa^ is estimated according to equation (33).

[0168] The mathematical model value QΨa^ for estimating armature flux linkage includes the d-axis inductance Ld and the armature flux linkage Ψa as motor parameters, as shown in equation (30). However, since the estimated d-axis inductance Ld^ converges to the true value according to equation (32), the estimated armature flux linkage Ψa^ also converges to the true value according to equation (33).

[0169] The q-axis inductance estimator 90 calculates the estimated q-axis inductance Lq^ by integrally controlling the difference (QLq - QLq^) between the reference model value QLq for q-axis inductance estimation output from the subtractor 78 and the mathematical model value QLq^ for q-axis inductance estimation, using the detected q-axis current value iq and the detected electrical angular velocity value ωe. Specifically, the estimated q-axis inductance Lq^ is estimated according to equation (34).

[0170] The mathematical model value QLq^ for estimating the q-axis inductance includes the d-axis inductance Ld, the armature flux linkage Ψa, and the q-axis inductance Lq as motor parameters, as shown in equation (31). However, since the estimated d-axis inductance Ld^ converges to the true value by equation (32) and the estimated armature flux linkage Ψa^ converges to the true value by equation (33), the estimated q-axis inductance Lq^ also converges to the true value by equation (34).

[0171] In equations (32), (33), and (34), "G" is the estimation gain used to determine the convergence response in parameter estimation. The estimation gain G can be set, for example, using an arbitrary time constant τ [sec] as shown in equation (35).

[0172] The same time constant τ may be used for parameter estimation in equations (32), (33), and (34). Alternatively, different time constants may be set if there is mutual interference during the parameter convergence process. This allows for variations in the time it takes for each motor parameter to converge, reducing the effects of mutual interference.

[0173] The estimated d-axis inductance Ld^, estimated armature flux linkage Ψa^, and estimated q-axis inductance Lq^ calculated in this manner are output to the MTPA current command value calculator 62 and the decoupling controller 69 and used as motor parameters.

[0174] The MTPA current command value calculator 62 (current command value generator 40) performs maximum torque / current control for the motor M based on the estimated values ​​of these multiple motor parameters. This makes it possible to represent the MTPA curve with appropriate motor parameters, and for example, it becomes possible to correct the d-axis current command value id_mtpa* (MTPA operating point) on the MTPA curve calculated according to equation (2). As a result, it becomes possible to drive the motor M according to the current motor parameters, thereby suppressing copper loss and achieving efficient motor control.

[0175] Furthermore, the decoupling controller 69 can calculate the d-axis decoupling correction value Vd_ff and the q-axis decoupling correction value Vq_ff using appropriate motor parameters (see equations (6) and (7)). This makes it possible to accurately correct the voltage command values ​​for the d-axis and q-axis, thereby achieving stable motor control.

[0176] Furthermore, the parameter estimation method according to this embodiment can be executed continuously while the motor M is being driven. This allows for seamless correction of multiple motor parameters and enables consistently efficient motor control. However, depending on the conditions, noise due to the superposition of high frequencies (Δidh) and high-frequency power loss may occur. For this reason, for example, the parameter estimation method according to this embodiment may be executed only when there is a change in the load current.

[0177] [Simulation Results] Here, the parameter estimation method of the comparative example and the parameter estimation method according to this embodiment are compared by simulation regarding the accuracy of motor parameter identification. Figure 7 is a table showing the motor parameters and operating conditions used in the simulation.

[0178] The motor M had 6 poles, and the armature winding resistance Ra was set to 0.6495 [Ω]. The d-axis inductance Ld was set to 9.3 [mH], the q-axis inductance Lq to 12.6 [mH], and the armature flux linkage Ψa to 0.15056 [Wb]. The values ​​of Ld, Lq, and Ψa shown in Figure 7 are the true values ​​that should converge through the estimation process.

[0179] The operating conditions were set to a machine angular rotation speed of 50 [rpm], a d-axis current id to -1.0 [A], and a q-axis current iq to 6.0 [A]. The d-axis current id and q-axis current iq shown in Figure 7 correspond to the detected values ​​input to the parameter estimator 51. Furthermore, the parameters for the d-axis current high frequency Δidh were set to a high frequency angular frequency ωh of 400π [rad / s] and a high frequency amplitude idh to 1.0 [A].

[0180] First, the identification characteristics of each motor parameter in the parameter estimation method of the comparative example will be explained. Figures 8, 9, and 10 are graphs showing the identification characteristics of armature linkage flux Ψa, d-axis inductance Ld, and q-axis inductance Lq according to the parameter estimation method of the comparative example. The upper graphs in Figures 8, 9, and 10 plot the time evolution of the mathematical model values ​​(QΨa^, QLd^, QLq^), and the lower graphs plot the time evolution of the parameter estimate values ​​(Ψa^, Ld^, Lq^). The horizontal axis of each graph is time [msec]. Note that each mathematical model value is expressed in units of instantaneous reactive power Q [Var].

[0181] In the parameter estimation method of the comparative example, the control is performed to set id = 0 in the estimation process of the armature flux linkage Ψa shown in Figure 8. Also, the estimation process of the d-axis inductance Ld shown in Figure 9 is started using the converged value after the estimated armature flux linkage Ψa^ in Figure 8 has converged. Here, the convergence of the estimated armature flux linkage Ψa^ means, for example, that the mathematical model value QΨa^ substantially matches the normative model value QΨa (for example, that the difference between the two (QΨa - QΨa^) has become smaller than a predetermined threshold). Similarly, the estimation process of the q-axis inductance Lq shown in Figure 10 is started using the converged value of the estimated armature flux linkage Ψa^ and the converged value of the estimated d-axis inductance Ld^ after the d-axis inductance estimate Ld^ in Figure 9 has converged.

[0182] Thus, in the parameter estimation method of the comparative example, the estimation process for each motor parameter is executed sequentially. For example, if it takes about 300 msec for one motor parameter to converge, then it would simply take 300 msec × 3 = 900 msec for all motor parameters to converge.

[0183] Furthermore, as shown in Figures 8, 9, and 10, in the parameter estimation method of the comparative example, although the mathematical model values ​​of each motor parameter converge to match the normative model values, the estimated values ​​of each motor parameter do not converge to the true values. That is, the estimated armature flux linkage Ψa^, the estimated d-axis inductance Ld^, and the estimated q-axis inductance Lq^ do not converge to the armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq set in Figure 7. This is because a steady-state model that does not include the p-term voltage is used. In the example shown here, the errors from the true values ​​were 11.5% for armature flux linkage, -11.6% for d-axis inductance, and 4.1% for q-axis inductance.

[0184] Next, the identification characteristics of each motor parameter in the parameter estimation method according to this embodiment will be described. Figures 11, 12, and 13 are graphs showing the identification characteristics of the d-axis inductance Ld, armature linkage flux Ψa, and q-axis inductance Lq according to the parameter estimation method according to this embodiment.

[0185] In the parameter estimation method according to this embodiment, the estimation processes for the d-axis inductance Ld, armature linkage flux Ψa, and q-axis inductance Lq are performed simultaneously without the need to control id=0. In other words, the processes shown in Figures 11, 12, and 13 can be started simultaneously at time 0 without restricting the operating conditions of the motor M.

[0186] For example, the d-axis inductance Ld shown in Figure 11 converges to the estimated d-axis inductance Ld^. At this time, the armature flux linkage Ψa shown in Figure 12 uses Ld^ shown in Figure 11, and the estimated armature flux linkage Ψa^ converges together with Ld^. Furthermore, the q-axis inductance Lq shown in Figure 13 uses Ld^ shown in Figure 11 and Ψa^ shown in Figure 12, and the estimated q-axis inductance Lq^ converges together with Ld^ and Ψa^.

[0187] As described above, in the parameter estimation method according to this embodiment, the estimation process for each motor parameter is executed in parallel. Therefore, compared to a method that estimates three parameters sequentially, as in the comparative example, the time required to estimate all motor parameters can be reduced to about one-third.

[0188] Furthermore, as shown in Figures 11, 12, and 13, in the parameter estimation method according to this embodiment, the mathematical model value of each motor parameter converges to match the reference model value, and consequently, the estimated value of each motor parameter converges to the true value. That is, the estimated armature flux linkage Ψa^, the estimated d-axis inductance Ld^, and the estimated q-axis inductance Lq^ converge to the armature flux linkage Ψa, d-axis inductance Ld, and q-axis inductance Lq set in Figure 7, respectively. This is because a transient model including the p-term voltage is used. Thus, the parameter estimation method according to this embodiment makes it possible to estimate motor parameters with greater accuracy compared to the comparative example.

[0189] As described above, in the motor control device 30 according to this embodiment, a plurality of corresponding normative model values ​​(QLd, QΨa, QLq) and a plurality of mathematical model values ​​(QLd^, QΨa^, QLq^) are calculated according to a transient model representing the instantaneous reactive power Q generated by superimposing the high-frequency command value Δidh on the d-axis current command value id_mtpa*, and a plurality of motor parameters are estimated using these model values. In addition, a component containing only a single motor parameter is calculated as one of the plurality of normative model values ​​in the transient model. In this embodiment, the normative model value QLd for estimating the d-axis inductance is calculated as a component containing only a single motor parameter. Furthermore, it becomes possible to independently estimate a single motor parameter (in this case, the d-axis inductance Ld). In addition, it becomes possible to perform estimation processing of other motor parameters in parallel with the estimation processing of a single motor parameter, while sequentially reflecting the estimation result. This makes it possible to estimate motor parameters in a shorter time over a wider operating range.

[0190] Maximum Torque / Current (MTPA) control, used in motor vector control, optimizes the current phase and effectively utilizes not only the magnet torque but also the reluctance torque to output the desired torque with the minimum current amplitude. However, in MTPA control, the MTPA operating point may deviate from the optimal value due to a decrease in inductance caused by magnetic saturation or fluctuations in armature flux linkage due to temperature changes. This increases copper loss and worsens efficiency, so it is desirable to adjust the MTPA operating point according to the operating conditions.

[0191] For example, if motor parameters such as inductance and armature flux linkage can be identified online, it becomes possible to modify the MTPA operating point. One method for identifying motor parameters online is the parameter estimation method described in the comparative example above. However, the parameter estimation method in the comparative example is time-consuming to estimate motor parameters and requires control to set id=0, which limits the applicable operating range. Furthermore, because it uses a model that does not include the p-term voltage, there is a problem with the accuracy of motor parameter identification.

[0192] In contrast, in this embodiment, by setting a component of the instantaneous reactive power Q that includes a single motor parameter as a reference model value and a mathematical model value, at least one motor parameter (in this case, the d-axis inductance Ld) can be independently estimated without performing control such as setting id = 0. This makes it possible to simultaneously identify the armature flux linkage Ψa, the d-axis inductance Ld, and the q-axis inductance Lq without limiting the operating range.

[0193] Furthermore, in this embodiment, a transient model of instantaneous reactive power Q is constructed using the voltage equation of the IPM motor in a transient state. This allows for the estimation of each motor parameter, including the p-term voltage generated by the high-frequency Δidh superimposed on the d-axis current, thereby improving the accuracy of motor parameter identification. As a result, it becomes possible to optimize the MTPA operating point, minimizing copper loss in the motor M and achieving efficient motor control.

[0194] <Second Embodiment> A motor control device according to a second embodiment of the present invention will now be described. In the following description, parts that are similar to the configuration and operation of the motor control device 30 described in the above embodiment will be omitted or simplified.

[0195] Figure 14 is a block diagram showing an example configuration of a motor control device according to the second embodiment. The motor control device 230 shown in Figure 14 has the addition of a notch filter processor 95, a high-frequency tracking controller 96, and a p-term voltage calculator 97 compared to the motor control device 30 shown in Figure 2.

[0196] Here, we will explain the parameters of the d-axis current high-frequency command value Δidh. The estimation accuracy of the motor parameters depends on the high-frequency amplitude idh and angular frequency ωh of the d-axis current high-frequency command value Δidh. For example, increasing the high-frequency amplitude idh or angular frequency ωh increases the values ​​of the mathematical model value and the normative model value, as shown in equations (29), (30), and (31). This makes it possible to extract the normative model value accurately from, for example, the instantaneous reactive power Q, thereby improving the estimation accuracy of the motor parameters.

[0197] In this regard, for example, increasing the high-frequency amplitude idh is undesirable because it raises concerns about increased power consumption and increased speed fluctuations due to variations in reluctance torque. On the other hand, increasing the angular frequency ωh may worsen the current tracking performance of the high-frequency component depending on the response characteristics of the current control. As a result, problems may arise such as the d-axis current id not tracking the d-axis current high-frequency command value Δidh, or the decoupling control not functioning properly, causing fluctuations in the q-axis current iq.

[0198] For example, in the above embodiment, the explanation was based on the assumption that the q-axis current iq is DC and the d-axis current id follows the d-axis current high-frequency command value Δidh. However, it is conceivable that increasing the angular frequency ωh may worsen the current tracking performance as described above. In this case, errors occur in the convergence values ​​of the motor parameters, leading to problems such as the inability to optimize the MTPA operating point.

[0199] To solve these problems, this embodiment provides the notch filter processor 95, high-frequency tracking controller 96, and p-term voltage calculator 97 described above. At least one of these three functional blocks may be provided.

[0200] The notch filter processor 95 is configured using a notch filter (band-stop filter) that attenuates components in a predetermined bandwidth. The notch filter processor 95 removes the fluctuating component corresponding to the d-axis current high-frequency command value Δidh from the estimated value of the rotational speed of the motor M (mechanical angular velocity detection value ωm) that is fed back to the current command value generator 40 (subtractor 60). In this embodiment, the notch filter processor 95 corresponds to a fluctuating component removal processor.

[0201] Specifically, the notch filter processor 95 is configured to attenuate the angular frequency ωh fluctuation component. For example, a velocity fluctuation component of angular frequency ωh is generated by the reluctance torque fluctuation that occurs when a high-frequency command value Δidh is superimposed on the d-axis current id. The notch filter processor 95 calculates a filtered mechanical angular velocity detection value ωm_n by attenuating the angular frequency ωh fluctuation component mentioned above from the mechanical angular velocity detection value ωm output from the 1 / Pn processor 52, and outputs it to the subtractor 60.

[0202] This makes it possible to suppress fluctuations in the q-axis current command value iq* and the d-axis current command value id_mtpa* caused by the speed controller 61 reacting to fluctuations in the angular frequency ωh. In addition, it becomes possible to set a higher frequency for the d-axis current high-frequency command value Δidh, which improves the signal-to-noise ratio in the motor parameter estimation process.

[0203] Furthermore, the method for removing the velocity fluctuation component at angular frequency ωh from the detected mechanical angular velocity value ωm is not limited to a notch filter. For example, a BPF (Ballpoint-Pass Filter) or the like may be used to extract the velocity fluctuation component at angular frequency ωh, and the system may be configured to subtract that component.

[0204] The high-frequency tracking controller 96 is installed in parallel with the current controller 67 and adjusts the voltage command values ​​(in this case, the d-axis voltage PI command value Vd_pi and the q-axis voltage PI command value Vq_pi) output from the current controller 67 so that the motor current flowing through the motor M follows the d-axis current high-frequency command value Δidh.

[0205] As explained with reference to Figure 2, the current controller 67 is composed of, for example, a PI controller. In this case, the gain of PI control is low for high-frequency currents, and it may be difficult to generate a motor current that properly follows the d-axis current high-frequency command value Δidh using only the output from the current controller 67. For this reason, the high-frequency tracking controller 96 compensates the output from the current controller 67 so that high frequencies are efficiently injected into the motor current.

[0206] The high-frequency tracking controller 96 calculates a d-axis current tracking compensation value Vd_h and a q-axis current tracking compensation value Vq_h to make the motor current track the d-axis current high-frequency command value Δidh, based on the d-axis current deviation (id* - id) output from the subtractor 65 and the q-axis current deviation (id* - id) output from the subtractor 66. The high-frequency tracking controller 96 also outputs the d-axis current tracking compensation value Vd_h to the adder 70 and the q-axis current tracking compensation value Vq_h to the adder 71.

[0207] As the high-frequency tracking controller 96, for example, a resonant regulator that resonates at the angular frequency ωh of the d-axis current high-frequency command value Δidh can be used. Alternatively, for example, it is possible to use repetitive control with a signal having an angular frequency ωh as the target signal. This makes it possible to generate a voltage command value that appropriately reflects the d-axis current high-frequency command value Δidh. Furthermore, it becomes possible to set a higher frequency for the d-axis current high-frequency command value Δidh, thereby improving the signal-to-noise ratio in the motor parameter estimation process.

[0208] The p-term voltage calculator 97 calculates the p-term voltage generated by superimposing the d-axis current high-frequency command value Δidh onto the d-axis current. Specifically, the p-term voltage calculator 97 calculates an orthogonal component that is generated in accordance with the time change of the current due to the d-axis current high-frequency command value Δidh and is orthogonal to the d-axis current high-frequency command value Δidh. This orthogonal component becomes the p-term voltage. In this embodiment, the p-term voltage calculator 97 corresponds to an orthogonal component calculator.

[0209] The p-term voltage calculator 97 calculates the p-term voltage pLdid (orthogonal component) according to equation (36) from the amplitude idh and angular frequency ωh of the d-axis current high-frequency command value Δidh output from the high-frequency command value generator 41. Here, Δidh is assumed to be a cosine wave as shown in equation (3). Note that equation (36) is the same as equation (19) described in the explanation of the transient model.

[0210] In this case, the decoupling controller 69 performs decoupling control on the voltage command value based on the orthogonal component (the p-term voltage shown in equation (36)) output from the p-term voltage calculator 97. Specifically, the decoupling controller 69 adds the p-term voltage pLdid to the d-axis decoupling correction value Vd_ff shown in equation (6) according to equation (37).

[0211] Furthermore, the q-axis decoupling correction value Vq_ff can be calculated using equation (7) in the same way as in the above embodiment, since there is no fluctuation in the q-axis current. In this way, by using the p-term voltage (orthogonal component), highly accurate decoupling control becomes possible. In addition, it becomes possible to set a higher frequency for the d-axis current high-frequency command value Δidh, which improves the signal-to-noise ratio in the motor parameter estimation process.

[0212] In this embodiment, the method described above enables the conversion of the q-axis current command value iq* to DC and improves the tracking ability with respect to the d-axis current high-frequency command value Δidh. As a result, increasing the angular frequency ωh improves the signal-to-noise ratio of the motor parameter estimation process, thereby improving the accuracy of motor parameter estimation. Consequently, the MTPA operating point is optimized, and copper loss can be minimized.

[0213] <Other Embodiments> The present invention is not limited to the embodiments described above, and various other embodiments can be realized.

[0214] In the above embodiment, in the transient model of instantaneous reactive power Q shown in equation (21), we focused on the fifth term of equation (21), which includes only the d-axis inductance Ld as a single-parameter component including a single motor parameter, and set the normative model value QLd for d-axis inductance estimation and the mathematical model value QLd^ for d-axis inductance estimation. However, we are not limited to this, and other terms including a single motor parameter can also be used as single-parameter components.

[0215] For example, the fourth term of equation (21) (ωe(Ld / 2)idh 2cos(2ωht) is a single-parameter component that includes only the d-axis inductance Ld. This component is an oscillatory component of cos(2ωht) with an angular frequency twice the angular frequency ωh of the d-axis current high-frequency command value Δidh, and can be extracted, for example, using a BPF that allows oscillatory components with angular frequency 2ωh to pass through.

[0216] Therefore, for example, by extracting the oscillation component with angular frequency 2ωh from the instantaneous reactive power Q (see equation (25)) calculated by the instantaneous reactive power calculator 80 using a BPF, that component can be used as the reference model value QLd for estimating the d-axis inductance. Furthermore, by using the estimated d-axis inductance value QLd^ as the d-axis inductance Ld in the fourth term of equation (21), the mathematical model value QLd^ for estimating the d-axis inductance can be calculated. Note that for other motor parameters (americ flux linkage Ψa and q-axis inductance Lq), the reference model value and mathematical model value are calculated in the same manner as in the above embodiment.

[0217] Thus, even when the reference model value QLd and the mathematical model value QLd^ for d-axis inductance estimation are calculated based on the fourth term of equation (21), it becomes possible to estimate the d-axis inductance Ld independently. This makes it possible to simultaneously estimate the d-axis inductance Ld, armature flux linkage Ψa, and q-axis inductance Lq without performing control such as setting id=0, enabling the estimation of motor parameters in a shorter time over a wider operating range.

[0218] Furthermore, the above embodiment described a method for estimating each motor parameter based on a transient model of instantaneous reactive power Q. However, the present invention is not limited to this and can also be applied to a method for estimating motor parameters using a steady-state model of instantaneous reactive power Q.

[0219] For example, the fourth term of the steady-state model of instantaneous reactive power Q shown in equation (12) is the same as the fourth term of equation (21), and is a single parameter component that includes only the d-axis inductance Ld. Therefore, even when using the steady-state model of instantaneous reactive power Q, the reference model value QLd for d-axis inductance estimation and the mathematical model value QLd^ for d-axis inductance estimation can be calculated based on the fourth term of equation (12). As a result, unlike the parameter estimation method of the comparative example described above, the d-axis inductance Ld can be estimated independently without performing control to set id=0.

[0220] In this case, the reference model value QΨa and the mathematical model value QΨa^ for estimating armature flux linkage can be calculated from the third term of equation (12), and the armature flux linkage Ψa can be estimated. Furthermore, the reference model value QLq and the mathematical model value QLq^ for estimating q-axis inductance can be calculated from the first and second terms (DC components) of equation (12), and the q-axis inductance Lq can be estimated.

[0221] Thus, even when using a steady-state model of instantaneous reactive power Q, it becomes possible to simultaneously estimate the d-axis inductance Ld, armature flux linkage Ψa, and q-axis inductance Lq without performing control such as setting id=0. For example, this method can be used when it is possible to estimate each motor parameter with the desired accuracy even when using a steady-state model. This makes it possible to broaden the operating range in which motor parameters can be estimated and shorten the time required for motor parameter estimation, compared to, for example, the parameter estimation method of the comparative example.

[0222] It is also possible to combine at least two of the feature features of the present invention described above. In other words, the various feature features described in each embodiment may be combined arbitrarily without distinction between embodiments. Furthermore, the various effects described above are merely examples and are not limiting, and other effects may also be exhibited.

[0223] M...Motor 20...Compressor 30, 230...Motor control device 40...Current command value generator 41...High frequency command value generator 42...Voltage command value generator 51...Parameter estimator 74...Instantaneous reactive power reference model value calculator 75...Instantaneous reactive power mathematical model value calculator 79...Integrator for parameter estimation 95...Notch filter processor 96...High frequency tracking controller 97...p-term voltage calculator 100...Air conditioner

Claims

1. A motor control device for estimating multiple motor parameters relating to a motor, comprising: a current command value generator that generates a q-axis current command value and a d-axis current command value based on a motor rotation speed command value; a high-frequency command value generator that generates a high-frequency command value superimposed on the d-axis current command value; a voltage command value generator that generates a voltage command value for driving the motor based on the q-axis current command value and the d-axis current command value with the high-frequency command value superimposed on it; a motor current detector that detects the motor current flowing through the motor; a normative model value calculator that calculates instantaneous reactive power by superimposing the high-frequency command value on the d-axis current command value based on the voltage command value and the motor current, and calculates a plurality of normative model values ​​for estimating each of the plurality of motor parameters from the calculation result of the instantaneous reactive power according to a predetermined model representing the instantaneous reactive power; a mathematical model value calculator that calculates a plurality of mathematical model values ​​corresponding to the plurality of normative model values ​​using the plurality of motor parameters according to the predetermined model; and an estimation processor that estimates the plurality of motor parameters based on the plurality of normative model values ​​and the plurality of mathematical model values, wherein the parameter estimator comprises: The aforementioned normative model value calculator is a motor control device that calculates, as one of the plurality of normative model values, a component that includes only a single motor parameter in the predetermined model from at least the calculation result of the instantaneous reactive power.

2. A motor control device according to claim 1, wherein the estimation processor performs a plurality of estimation processes for estimating the plurality of motor parameters in parallel.

3. A motor control device according to claim 1, wherein the predetermined model is a transient model of instantaneous reactive power that includes an in-phase component of the high-frequency command value and an orthogonal component that is generated in accordance with the time change of the current due to the high-frequency command value and is orthogonal to the high-frequency command value.

4. A motor control device according to claim 3, wherein the plurality of motor parameters include armature flux linkage, d-axis inductance, and q-axis inductance, the transient model includes an orthogonal component to the high-frequency command value as a component including only the d-axis inductance, the reference model value calculator comprises an orthogonal component separator that separates the calculation result of the instantaneous reactive power into an orthogonal component and an in-phase component to the high-frequency command value, calculates the orthogonal component as a first reference model value used for estimating the d-axis inductance, and calculates the in-phase component as a second reference model value used for estimating the armature flux linkage, and a low-pass filter that calculates the DC component included in the calculation result of the instantaneous reactive power as a third reference model value used for estimating the q-axis inductance.

5. A motor control device according to claim 4, wherein the estimation processor estimates the motor parameters by integrally controlling the difference between the corresponding normative model values ​​and the mathematical model values.

6. A motor control device according to claim 1, wherein the current command value generator performs maximum torque / current control for the motor based on the estimated values ​​of the plurality of motor parameters estimated by the parameter estimator.

7. A motor control device according to claim 6, further comprising a fluctuation component removal processor that removes a fluctuation component corresponding to the high-frequency command value from the estimated value of the rotational speed of the motor that is fed back to the current command value generator.

8. A motor control device according to claim 6, further comprising a high-frequency tracking controller that adjusts the voltage command value so that the motor current follows the high-frequency command value.

9. A motor control device according to claim 6, further comprising: an orthogonal component calculator that calculates an orthogonal component that is generated in accordance with the time change of the current due to the high-frequency command value and is orthogonal to the high-frequency command value; and a decoupling controller that performs decoupling control with respect to the voltage command value based on the orthogonal component.

10. An air conditioner equipped with a motor control device according to any one of claims 1 to 9, comprising: a motor; a refrigerant circuit including a compressor driven by the motor; and an inverter circuit that drives the motor according to a voltage command value.

Images