JP7818151B2 - Motor control device - Google Patents

Description

The present invention relates to a motor control device that controls a motor.

Conventionally, methods for controlling a motor by estimating rotor position without using a position sensor to detect the rotor position are known. For example, Patent Document 1 describes a motor control device for a permanent magnet synchronous motor that drives a compressor, which uses position sensorless vector control to perform torque control so that the motor torque follows the compressor load torque. In such a motor control device, an axis error is calculated based on the motor current, an estimated motor speed is calculated based on the axis error, and the rotor position is estimated based on the estimated motor speed.

Japanese Patent Application Laid-Open No. 2021-125898

When a motor is used to drive a compressor or other device, the load torque fluctuates periodically during one rotation of the rotor. When torque control or other control is performed on a motor connected to a load with such periodic load torque fluctuations, the motor current may also fluctuate periodically. If periodic fluctuations in motor current are not taken into account when estimating rotor position, the accuracy of the rotor position estimation may decrease.

For example, in the voltage saturation region where the motor output voltage is limited by the DC voltage supplied to the inverter that drives the motor, in addition to the periodic fluctuations in load torque described above, it is also necessary to suppress fluctuations in the amplitude of the voltage command value (output voltage amplitude). One way to suppress such fluctuations in both load torque and voltage amplitude is to adjust the motor current, but in this case, the periodic fluctuations in the motor current also become more pronounced, which can easily reduce the accuracy of rotor position estimation.

In light of the above circumstances, the object of the present invention is to provide a motor control device that can estimate rotor position with high accuracy even when the motor current fluctuates periodically.

In order to achieve the above object, a motor control device according to one aspect of the present invention includes a current detection unit, a current fluctuation extraction unit, and a position estimation unit.
The current detection unit detects a motor current flowing through a motor.
The current fluctuation extraction unit extracts a current fluctuation component that is a periodic fluctuation component of the motor current.
The position estimator estimates a rotor position of the motor based on the current fluctuation component.

In this motor control device, the motor rotor position is estimated based on the current fluctuation component, which is a periodic fluctuation component contained in the motor current. This makes it possible to estimate the rotor position with high accuracy even when the motor current fluctuates periodically.

The motor control device may further include an inductance voltage calculation unit that calculates an inductance voltage, which is an induced voltage generated by the current fluctuation component, based on the inductance of the motor and the differential value of the current fluctuation component. In this case, the position estimation unit may estimate the rotor position based on the inductance voltage.

This makes it possible to estimate the rotor position by taking into account the inductance voltage, which is an induced voltage generated by current fluctuation components, thereby improving the accuracy of rotor position estimation.

The current fluctuation extraction unit may extract two or more current fluctuation components having different periods. In this case, the inductance voltage calculation unit may calculate the inductance voltage for each of the two or more current fluctuation components based on the differential value of the current fluctuation component, and calculate a total inductance voltage by summing the inductance voltages for each of the two or more current fluctuation components. The position estimation unit may also estimate the rotor position based on the total inductance voltage.

This makes it possible to estimate the rotor position by taking into account the inductance voltage generated by each current fluctuation component with a different period, thereby further improving the accuracy of rotor position estimation.

The current fluctuation extraction unit may extract the current fluctuation component by expanding the motor current into a Fourier series.

By using Fourier series expansion, it is possible to extract the current fluctuation component with the influence of motor current noise suppressed, enabling the rotor position to be estimated with high accuracy.

The current fluctuation extraction unit may extract first-order current fluctuation components and second-order current fluctuation components of the motor current by expanding the motor current into a Fourier series.

For example, the more types of current fluctuation components used to estimate rotor position, the more accurate the rotor position estimation, but the longer the processing time. Therefore, by using only primary and secondary current fluctuation components, it is possible to achieve both high accuracy in position estimation and reduced processing time.

The position estimation unit may estimate the rotor position based on the current fluctuation component when performing torque control on the motor.

This makes it possible to estimate the rotor position with high accuracy when performing torque control.

The position estimation unit may estimate the rotor position based on the current fluctuation component when the control region of the motor is a voltage saturation region.

This makes it possible to estimate the rotor position with high accuracy in the voltage saturation region.

As described above, according to the present invention, it is possible to estimate the rotor position with high accuracy even when the motor current fluctuates periodically. Note that the effects described here are not necessarily limited to those described herein, and any of the effects described in this disclosure may be used.

10A to 10C are diagrams illustrating an example of the operation of the motor control device of the present disclosure. 10A to 10C are diagrams illustrating an example of the operation of the motor control device of the present disclosure. 1 is a diagram illustrating an example of the configuration of a motor control device according to an embodiment of the present invention; FIG. 4 is a diagram illustrating an example of the configuration of a control switch determination unit according to the present embodiment. FIG. 4 is a diagram illustrating an example of the configuration of a correction torque generator according to the present embodiment. FIG. 2 is a diagram illustrating an example of the configuration of a voltage saturation region voltage command value generator according to the present embodiment. FIG. 4 is a diagram illustrating an example of the configuration of an output voltage limit command value generator according to the present embodiment. 5A and 5B are diagrams illustrating an example of the operation of a current command value calculator according to the present embodiment. 5A and 5B are diagrams illustrating an example of the operation of a current command value calculator according to the present embodiment. 10A and 10B are diagrams illustrating an example of the operation of the voltage vector angle calculator according to the present embodiment. 10A and 10B are diagrams illustrating an example of the operation of the MTPI voltage amplitude limiting processor according to the present embodiment. FIG. 4 is a diagram illustrating an example of an output voltage waveform. 10 is a graph showing measurement data during motor control using a steady-state model given as a comparative example and a transient model according to the present embodiment. 10 is a graph showing measurement data during motor control using a steady-state model as a comparative example. 6 is a graph showing measurement data during motor control using a transient model according to the present embodiment.

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

This disclosure describes an example of a motor control device that uses position sensorless vector control to control the torque of a permanent magnet synchronous motor (PMSM) that drives a compressor with periodic load torque fluctuations, such as a motor control device used in an air conditioner or cryogenic storage device. However, the disclosed technology is widely applicable to motor control devices that perform torque control of a motor that drives a load with periodic load torque fluctuations.

<Operation of the motor control device>
1A and 1B are diagrams illustrating an example of the operation of the motor control device of the present disclosure.

The constant induced voltage ellipse shown in Figures 1A and 1B (only a portion of the ellipse is shown in Figures 1A and 1B) is a current vector locus where the induced voltage Vo of the motor is constant, and the diameter of the constant induced voltage ellipse becomes smaller as the estimated electrical angle angular velocity ωe increases. The constant induced voltage ellipse shown in Figure 1A represents the current vector locus when the estimated electrical angle angular velocity ωe is constant. The constant induced voltage ellipse shown in Figure 1B represents the current vector locus when the estimated electrical angle angular velocity ωe fluctuates due to load torque fluctuations. Figure 1B shows the constant induced voltage ellipse at the maximum value of the estimated electrical angle angular velocity ωe, the constant induced voltage ellipse at the minimum value of the estimated electrical angle angular velocity ωe, and the constant induced voltage ellipse at the average value of the estimated electrical angle angular velocity ωe.

When performing torque control in the voltage saturation region where flux-weakening control of the motor is performed, the motor control device of the present disclosure calculates the q-axis current command value Iq* and the d-axis current command value Id* based on the intersection of the constant torque curve T* (= To*±ΔT), which fluctuates by ±ΔT due to torque control, with the fluctuating constant induced voltage ellipse, as shown in FIG. 1B. For example, the motor control device of the present disclosure calculates the d-axis current command value Id* and the q-axis current command value Iq* based on the intersection of the constant torque curve, which is the current locus at which the total torque command value T*, calculated by adding the fluctuating torque command value ΔT, which is the correction torque, to the average torque command value To*, is constant, with the constant induced voltage ellipse, which is the current locus at which the induced voltage command value V0* and the estimated electrical angle angular velocity ωe, which are constant, are required to achieve the desired output voltage amplitude Va (the amplitude of the voltage command value). Note that the constant induced voltage ellipse and the constant torque curve are not uniquely determined based on motor parameters such as reactance, but change from moment to moment depending on the operating state of the motor.

Furthermore, since the motor to be controlled in this disclosure drives a compressor whose load torque fluctuates periodically, it is expected that the motor's rotation speed in terms of mechanical angle (estimated mechanical angle angular velocity ωm) will contain first-order fluctuation components that fluctuate with the load torque fluctuation period, as well as higher-order fluctuation components with periods that are integer multiples of the load torque fluctuation period. Therefore, the mechanical angle phase θm and motor current (d-axis current Id and q-axis current Iq) also contain first-order fluctuation components and higher-order fluctuation components due to the periodic load torque fluctuation. The motor control device will perform the torque control described above in a state in which the mechanical angle phase θm and motor current also contain periodic fluctuation components.

<Configuration of the motor control device>
Fig. 2 is a diagram showing an example of the configuration of a motor control device according to this embodiment. In Fig. 2, motor control device 100 includes subtractors 11 and 38, a speed controller 12, an adder 13, a voltage command value generator 14, a control switch determination unit 15, a d-q/u, v, w converter 23, a PWM (Pulse Width Modulation) modulator 24, and an IPM (Intelligent Power Module) 25. IPM 25 is connected to motor M. An example of motor M is a PMSM.

The motor control device 100 also has a shunt resistor 26, current sensors 27a and 27b, and a 3φ current calculator 28. It is sufficient for the motor control device 100 to have either the shunt resistor 26 or the current sensors 27a and 27b.

The motor control device 100 also includes a u, v, w/d-q converter 29, a current fluctuation extractor 35, an inductance voltage calculator 36, an axis error calculator 30, a PLL (Phase Locked Loop) controller 31, a position estimator 32, a 1/Pn processor 33, and a correction torque generator 34.

The voltage command value generator 14 includes a normal control region voltage command value generator 14a, a voltage saturation region voltage command value generator 14b, a switch SW1, and a switch SW2. Switch SW1 includes contacts 14c-1, 14c-2, and 14c-3. Switch SW2 includes contacts 14c-4, 14c-5, and 14c-6.

The subtractor 11 calculates the angular velocity error Δω by subtracting the mechanical angular velocity estimated ωm, which is the current estimated angular velocity output from the 1/Pn processor 33, from the mechanical angular velocity command value ωm* input to the motor control device 100 from outside the motor control device 100 (for example, from a higher-level controller), and outputs the calculated angular velocity error Δω to the speed controller 12.

The speed controller 12 generates an average torque command value To* such that the angular velocity error Δω input from the subtractor 11 approaches zero, and outputs the generated average torque command value To* to the adder 13.

The adder 13 calculates the total torque command value T* by adding the average torque command value To* output from the speed controller 12 and the fluctuating torque command value ΔT output from the correction torque generator 34, and outputs the calculated total torque command value T* to the voltage command value generator 14.

The voltage command value generator 14 generates a d-axis voltage command value Vd* and a q-axis voltage command value Vq* based on the total torque command value T* output by the adder 13 in each of the normal control region and the voltage saturation region, and outputs the generated d-axis voltage command value Vd* and q-axis voltage command value Vq*. The voltage saturation region is a region in which the output voltage amplitude Va saturates in the high rotation region of the motor M, and flux-weakening control is performed. The normal control region is a region outside the voltage saturation region, in which the motor M is controlled by varying the output voltage, and in the normal control region, maximum torque/current control, etc., is performed.

When the control switch determination unit 15 outputs a control signal CONTROL_TYPE: A (normal control), the voltage command value generator 14 connects the contact 14c-1 and the contact 14c-3 of the switch SW1 and also connects the contact 14c-4 and the contact 14c-6 of the switch SW2, and outputs the d-axis voltage command value Vd* and the q-axis voltage command value Vq* generated by the normal control region voltage command value generator 14a to the d-q/u, v, w converter 23. On the other hand, when the control switch determination unit 15 outputs a control signal CONTROL_TYPE: B (voltage saturation control), the voltage command value generator 14 connects the contact 14c-2 and the contact 14c-3 of the switch SW1 and also connects the contact 14c-5 and the contact 14c-6 of the switch SW2, and outputs the d-axis voltage command value Vd* and the q-axis voltage command value Vq* generated by the voltage saturation region voltage command value generator 14b to the d-q/u, v, w converter 23.
, v, w converter 23.

The control switching determination unit 15 determines whether the current control region of the motor M is the normal control region or the voltage saturation region based on the output voltage limit value Vdq_limit, the d-axis voltage command value Vd*, and the q-axis voltage command value Vq*. If the control switching determination unit 15 determines that the current control region of the motor M is the normal control region, it outputs a control signal CONTROL_TYPE: A (normal control) to the voltage command value generator 14. If the control switching determination unit 15 determines that the current control region of the motor M is the voltage saturation region, it outputs a control signal CONTROL_TYPE: B (voltage saturation control) to the voltage command value generator 14. The output voltage limit value Vdq_limit is obtained by converting the DC voltage Vdc supplied to the IPM 25 from outside the IPM 25 (for example, a power converter, not shown) into a voltage value in the dq rotating coordinate system, which is the control system.

The d-q/u,v,w converter 23 converts the two-phase d-axis voltage command value Vd* and q-axis voltage command value Vq* output from the voltage command value generator 14 into a three-phase U-phase output voltage command value Vu*, a V-phase output voltage command value Vv*, and a W-phase output voltage command value Vw* based on the electrical angle phase (dq-axis phase) θe, which is the current rotor position output from the position estimator 32. The dq/u,v,w converter 23 then outputs the U-phase output voltage command value Vu*, the V-phase output voltage command value Vv*, and the W-phase output voltage command value Vw* to the PWM modulator 24.

The PWM modulator 24 generates six-phase PWM signals based on the U-phase output voltage command value Vu*, the V-phase output voltage command value Vv*, the W-phase output voltage command value Vw*, and the PWM carrier signal, and outputs the generated six-phase PWM signals to the IPM 25.

The IPM 25 converts the DC voltage Vdc supplied from outside the IPM 25 based on the six-phase PWM signal output from the PWM modulator 24 to generate AC voltages to be applied to the U, V, and W phases of the motor M, and applies each of these AC voltages to the U, V, and W phases of the motor M.

When the bus current is detected using the single-shunt method with shunt resistor 26, 3φ current calculator 28 calculates the U-phase current value Iu, V-phase current value Iv, and W-phase current value Iw of motor M from the six-phase PWM switching information output from PWM modulator 24 and the detected bus current. Alternatively, when U-phase current and V-phase current are detected by current sensors 27a and 27b, 3φ current calculator 28 calculates the remaining W-phase current value Iw based on Kirchhoff's law: "Iu + Iv + Iw = 0." 3φ current calculator 28 outputs the calculated phase current values Iu, Iv, and Iw of each phase to u, v, w/d-q converter 29.

The u,v,w/d-q converter 29 converts the three-phase U-phase current value Iu, V-phase current value Iv, and W-phase current value Iw output from the 3φ current calculator 28 into two-phase d-axis current Id and q-axis current Iq based on the electrical angle phase θe indicating the current rotor position output from the position estimator 32. The u,v,w/d-q converter 29 then outputs the d-axis current Id and q-axis current Iq to the voltage command value generator 14, current fluctuation extractor 35, and axis error calculator 30.

The U-phase current value Iu, V-phase current value Iv, and W-phase current value Iw of motor M are detected values obtained by detecting the current values of each phase of motor M. Therefore, the d-axis current Id and q-axis current Iq obtained by converting the U-phase current value Iu, V-phase current value Iv, and W-phase current value Iw can also be considered detected values. In vector control of motor M, the d-axis current Id and q-axis current Iq are used as detected values of the motor current flowing through motor M. In this embodiment, the 3φ current calculator 28 and u, v, w/d-q converter 29 constitute a current detection unit that detects the motor current flowing through the motor.

The current fluctuation extractor 35 calculates the d-axis current fluctuation component ΔId, which is the periodic fluctuation component of the d-axis current Id, using the d-axis current Id output from the u,v,w/d-q converter 29 and the mechanical angle phase θmn output from the position estimator 32. It also calculates the q-axis current fluctuation component ΔIq, which is the periodic fluctuation component of the q-axis current Iq, using the q-axis current Iq output from the u,v,w/d-q converter 29 and the mechanical angle phase θmn output from the position estimator 32.

Here, the mechanical angle phase θmn is the nth-order mechanical angle phase. In this embodiment, the first-order mechanical angle phase θm1 and the second-order mechanical angle phase θm2 are used. In this embodiment, the current fluctuation extractor 35 calculates the d-axis current fluctuation component ΔId and the q-axis current fluctuation component ΔIq for each order n of the nth-order current fluctuation component. The current fluctuation extractor 35 then outputs the d-axis current fluctuation component ΔId and the q-axis current fluctuation component ΔIq to the inductance voltage calculator 36. The operation of the current fluctuation extractor 35 will be explained in detail later. In this embodiment, the current fluctuation extractor 35 corresponds to a current fluctuation extraction unit.

The inductance voltage calculator 36 calculates an inductance voltage pLdId (hereinafter sometimes referred to as the inductance voltage pLdId caused by d-axis current fluctuation), which is an induced voltage caused by the d-axis current fluctuation component ΔId, using the d-axis current fluctuation component ΔId output by the current fluctuation extractor 35, the d-axis inductance Ld of the motor M, and the mechanical angle phase θmn output by the position estimator 32. The inductance voltage calculator 36 also calculates an inductance voltage pLdIq (hereinafter sometimes referred to as the inductance voltage pLdIq caused by q-axis current fluctuation), which is an induced voltage caused by the q-axis current fluctuation component ΔIq, using the q-axis current fluctuation component ΔIq output by the current fluctuation extractor 35, the d-axis inductance Ld of the motor M, and the mechanical angle phase θmn output by the position estimator 32. Here, the "p" in the inductance voltage pLdId generated by d-axis current fluctuations and the inductance voltage pLdIq generated by q-axis current fluctuations represents a differential operator. The d-axis inductance Ld of the motor M is stored in the motor control device 100.

In addition, in this embodiment, the inductance voltage calculator 36 calculates the inductance voltage pLdId generated by the d-axis current fluctuation and the inductance voltage pLdIq generated by the q-axis current fluctuation for each order n of the nth-order current fluctuation component. The inductance voltage calculator 36 then outputs the inductance voltage pLdId generated by the d-axis current fluctuation and the inductance voltage pLdIq generated by the q-axis current fluctuation to the axis error calculator 30. The operation of the inductance voltage calculator 36 will be explained in detail later. In this embodiment, the inductance voltage calculator 36 corresponds to the inductance voltage calculation unit.

The axis error calculator 30 calculates the axis error Δθ (the difference between the estimated rotation axis and the actual rotation axis) using the d-axis voltage command value Vd* and q-axis voltage command value Vq* output from the voltage command value generator 14, the d-axis current Id and q-axis current Iq output from the u,v,w/d-q converter 29, the inductance voltage pLdId generated by d-axis current fluctuations and the inductance voltage pLdIq generated by q-axis current fluctuations output from the inductance voltage calculator 36, and the electrical angle estimated angular velocity ωe output from the PLL controller 31. The axis error calculator 30 then outputs the calculated axis error Δθ to the PLL controller 31. In this way, the motor control device 100 calculates the axis error Δθ using the inductance voltages (inductance voltage pLdId caused by d-axis current fluctuations and inductance voltage pLdIq caused by q-axis current fluctuations) that correspond to the current fluctuation components (d-axis current fluctuation component ΔId and q-axis current fluctuation component ΔIq). The specific operation of the axis error calculator 30 will be described in detail later.

The PLL controller 31 calculates the electrical angle estimated angular velocity ωe, which is the current estimated angular velocity, based on the axis error Δθ output from the axis error calculator 30, and outputs the calculated electrical angle estimated angular velocity ωe to the position estimator 32, 1/Pn processor 33, voltage command value generator 14, and axis error calculator 30.

The position estimator 32 estimates the electrical angle phase θe and the mechanical angle phase θm based on the electrical angle estimated angular velocity ωe output from the PLL controller 31. The position estimator 32 then outputs the estimated electrical angle phase θe to the d-q/u,v,w converter 23 and the u,v,w/d-q converter 29, and outputs the estimated mechanical angle phase θm to the voltage command value generator 14 and the correction torque generator 34. Note that the mechanical angle phase θm output to the voltage command value generator 14 and the correction torque generator 34 is the first-order mechanical angle phase θm1. Hereinafter, when the mechanical angle phase "θm" is simply referred to, it refers to the first-order mechanical angle phase "θm1".

The position estimator 32 also estimates the second-order mechanical angle phase θm2 based on the electrical angle estimated angular velocity ωe. The position estimator 32 then outputs the first-order mechanical angle phase θm1 (θm) and the second-order mechanical angle phase θm2 as the nth-order mechanical angle phase θmn to the current fluctuation extractor 35 and the inductance voltage calculator 36.

The 1/Pn processor 33 calculates the estimated mechanical angle angular velocity ωm by dividing the estimated electrical angle angular velocity ωe output from the PLL controller 31 by the number of pole pairs Pn of the motor M, and outputs the calculated estimated mechanical angle angular velocity ωm to the subtractors 11 and 38.

The electrical angle estimated angular velocity ωe output by the PLL controller 31 and the mechanical angle estimated angular velocity ωm output by the 1/Pn processor 33 are parameters that estimate the rotational speed of the rotor of the motor M. Furthermore, the electrical angle phase θe and mechanical angle phase θm output by the position estimator 32 are parameters that estimate the rotor position of the motor M. In this embodiment, the axis error calculator 30, PLL controller 31, position estimator 32, and 1/Pn processor 33 constitute a position estimation unit that estimates the rotor position of the motor based on the current fluctuation component.

The subtractor 38 calculates the mechanical angle estimated angular velocity fluctuation Δωm by subtracting the mechanical angular velocity command value ωm* from the mechanical angle estimated angular velocity ωm output from the 1/Pn processor 33, and outputs the calculated mechanical angle estimated angular velocity fluctuation Δωm to the correction torque generator 34.

The correction torque generator 34 generates a fluctuating torque command value ΔT for suppressing the mechanical angle estimated angular velocity fluctuation Δωm, which is a periodic speed fluctuation, to the speed fluctuation tolerance value |Δωm|* or less based on the speed fluctuation tolerance value |Δωm|*, which is the speed fluctuation range within which vibrations of the motor M are tolerable, the mechanical angle estimated angular velocity fluctuation Δωm output from the subtractor 38, and the mechanical angle phase θm output from the position estimator 32, and outputs the generated fluctuating torque command value ΔT to the adder 13. The speed fluctuation tolerance value |Δωm|* is stored in the motor control device 100. The mechanical angle estimated angular velocity fluctuation (speed fluctuation) Δωm differs from the value of the angular velocity error Δω only in sign, positive or negative.

<Configuration of control switching determination unit>
3 is a diagram showing an example of the configuration of the control switching determination unit 15 according to this embodiment. In FIG. 3, the control switching determination unit 15 has a voltage amplitude calculator 15a and a control switching determiner 15b, and determines whether the current control region of the motor M is the normal control region or the voltage saturation region as follows.

The voltage amplitude calculator 15a calculates the output voltage amplitude Va based on the d-axis voltage command value Vd* and the q-axis voltage command value Vq* output from the voltage command value generator 14 in accordance with equation (1).

The control switching determiner 15b compares the peak value of the output voltage amplitude Va calculated by the voltage amplitude calculator 15a with the output voltage limit value Vdq_limit.

If the peak value of the output voltage amplitude Va is less than the output voltage limit value Vdq_limit, the control switching determiner 15b determines that the current control region of the motor M is the normal control region and outputs the control signal CONTROL_TYPE: A to the voltage command value generator 14.

On the other hand, if the peak value of the output voltage amplitude Va is equal to or greater than the output voltage limit value Vdq_limit, the control switching determiner 15b determines that the current control region of the motor M is the voltage saturation region and outputs the control signal CONTROL_TYPE: B to the voltage command value generator 14.

<Configuration of correction torque generator>
4 is a diagram showing an example of the configuration of the correction torque generator according to this embodiment. In FIG. 4, the correction torque generator 34 includes a speed fluctuation component separator 34a, a speed fluctuation amplitude calculator 34b, a subtractor 34c, a correction torque amplitude calculator 34d, a speed fluctuation phase corrector 34e, an orthogonal component separator 34f, and a correction torque demodulator 34g.

The correction torque generator 34 adjusts the amplitude (correction torque amplitude |Δωm|) and phase of the fluctuation torque command value (correction torque) ΔT for each mechanical angle cycle so that the speed fluctuation amplitude |Δωm| is within the speed fluctuation tolerance |Δωm|*, within which vibration of the motor M will not cause problems in actual use.

The speed fluctuation component separator 34a separates the mechanical angle estimated angular velocity fluctuation Δωm into two Fourier coefficients ωsin (sin component) and ωcos (cos component), which are the fundamental wave components of Δωm, according to equations (2.1) and (2.2). By calculating the Fourier coefficient of the fundamental wave component of the mechanical angle estimated angular velocity fluctuation Δωm for each mechanical angle cycle, it is possible to eliminate harmonic components of the mechanical angle estimated angular velocity fluctuation Δωm and accurately extract the fundamental wave component of the mechanical angle estimated angular velocity fluctuation Δωm. ωsin and ωcos are values that are updated for each mechanical angle cycle.

The speed fluctuation amplitude calculator 34b calculates the speed fluctuation amplitude |Δωm| based on the Fourier coefficients ωsin and ωcos in accordance with equation (3). Because ωsin and ωcos are values that are updated every mechanical angle period, the speed fluctuation amplitude |Δωm| is also updated every mechanical angle period.

The subtractor 34c calculates the speed fluctuation error |Δωm|err by subtracting the speed fluctuation tolerance |Δωm|* from the speed fluctuation amplitude |Δωm| output by the speed fluctuation amplitude calculator 34b. The speed fluctuation tolerance |Δωm|* specifies the speed fluctuation amplitude |Δωm| within the allowable range of vibration of the motor M.

The correction torque amplitude calculator 34d adjusts the correction torque amplitude |Δω| for each mechanical angle cycle in accordance with the error between the speed fluctuation amplitude |Δωm| and the speed fluctuation allowable value |Δωm|*. For example, the correction torque amplitude calculator 34d calculates the correction torque amplitude |ΔT| by multiplying the speed fluctuation error |Δωm|err, which is the error between the speed fluctuation amplitude |Δωm| and the speed fluctuation allowable value |Δωm|*, by a correction gain k according to equation (4) and adding the multiplication result to |ΔT|_old. In equation (4), |ΔT|_old is the correction torque amplitude |ΔT| for the previous mechanical angle cycle. By appropriately setting the correction gain k, it is possible to prevent hunting of the speed fluctuation |Δω| at the boundary of the speed fluctuation allowable value |Δωm|* and vibrations that occur when the speed fluctuation |Δω| exceeds the speed fluctuation allowable value |Δωm|* due to a sudden change in load torque.

The speed fluctuation phase corrector 34e corrects the phase of the mechanical angle estimated angular velocity fluctuation Δωm acquired for each mechanical angle cycle. For example, the speed fluctuation phase corrector 34e multiplies each of the Fourier coefficients ωsin and ωcos by a correction gain k according to equations (5.1) and (5.2), and adds ωsin_i_old and ωcos_i_old to the respective multiplication results. ωsin_i_old in equation (5.1) is ωsin_i in the previous mechanical angle cycle, and ωcos_i_old in equation (5.2) is ωcos_i in the previous mechanical angle cycle. Then, the speed fluctuation phase corrector 34e calculates the arctangent of ωsin_i and ωcos_i as the speed fluctuation corrected phase φωi according to equation (5.3). This speed fluctuation correction phase φωi becomes the phase reference when torque control is performed, and the phase delayed by π/2 from this reference becomes the phase of the fluctuation torque command value ΔT (correction torque phase).

The orthogonal component separator 34f calculates the sine component (ωsin_i) and cosine component (ωcos_i) of the speed fluctuation correction phase φωi in accordance with the corrected torque amplitude |ΔT| and the speed fluctuation correction phase φωi in accordance with the equations (6.1) and (6.2). This process also serves to prevent divergence during phase correction by the calculation of the equations (5.1) and (5.2).

The correction torque demodulator 34g calculates the fluctuation torque command value ΔT in accordance with the equations (7.1) and (7.2) based on the sine component (ωsin_i) and cosine component (ωcos_i) of the speed fluctuation correction phase φωi. This process converts the speed fluctuation correction phase φωi into a correction torque phase that is delayed by π/2, and generates an instantaneous value of the fluctuation torque command value ΔT at the mechanical angle phase θm.

The correction torque demodulator 34g may calculate the instantaneous value of the fluctuating torque command value ΔT according to the equation (8) instead of the equations (7.1) and (7.2).

Then, the adder 13 calculates the total torque command value T* by adding the fluctuating torque command value ΔT to the average torque command value To* output from the speed controller 12 according to equation (9).

<Configuration of voltage saturation region voltage command value generator>
5 is a diagram showing an example of the configuration of a voltage saturation region voltage command value generator according to this embodiment. In FIG. 5, the voltage saturation region voltage command value generator 14b includes an output voltage limit command value generator 14b1, an induced voltage command value calculator 14b2, a current command value calculator 14b3, a temporary voltage command value calculator 14b4, a voltage vector angle calculator 14b5, and a voltage command value calculator 14b6.

Figure 6 is a diagram showing an example configuration of an output voltage limit command value generator according to this embodiment. In Figure 6, the output voltage limit command value generator 14b1 includes an MTPI current command value calculator 14b1-1, an MTPI voltage command value calculator 14b1-2, an MTPI voltage amplitude calculator 14b1-3, an average output voltage generator 14b1-4, an MTPI voltage fluctuation component extractor 14b1-5, adders 14b1-6 and 14b1-8, and an MTPI voltage amplitude limit processor 14b1-7.

The output voltage limit command value generator 14b1 generates an output voltage limit command value Va* based on the total torque command value T*, the estimated electrical angle angular velocity ωe, the output voltage limit value Vdq_limit, the d-axis current Id, the q-axis current Iq, and the mechanical angle phase θm. This output voltage limit command value Va* is a voltage that adjusts the amplitude of the output voltage fluctuation within a range up to the output voltage limit value Vdq_limit and matches the phase of the output voltage fluctuation with the phase of the output voltage fluctuation in the normal control region (MTPI control region).

6, an MTPI current command value calculator 14b1-1 calculates an MTPI assumed d-axis current command value Id_mtpi* and an MTPI assumed q-axis current command value Iq_mtpi*, which are the intersections of a constant torque curve, which is a current locus at which the total torque command value T* is constant, and an MTPI curve (maximum torque/current control curve). The intersections of the constant torque curve and the MTPI curve are calculated using, for example, the motor torque equation of equation (10) and equation (11), which is the d-axis current equation on the MTPI curve when the q-axis current is known.

By eliminating the d-axis current Id from equations (10) and (11), a quartic equation for the q-axis current Iq can be obtained as shown in equation (12).

By applying, for example, Newton's method to the quartic equation shown in equation (12), it is possible to derive a solution corresponding to the MTPI assumed q-axis current command value Iq_mtpi* at the intersection of the constant torque curve of the total torque command value T* and the MTPI curve.

The MTPI current command value calculator 14b1-1 calculates the MTPI assumed d-axis current command value Id_mtpi* in accordance with the d-axis current equation (11) based on the MTPI assumed q-axis current command value Iq_mtpi*, which is the solution of equation (12).

The MTPI voltage command value calculator 14b1-2 calculates the MTPI assumed d-axis voltage Vd_mtpi* and the MTPI assumed q-axis voltage Vq_mtpi* based on the MTPI assumed d-axis current command value Id_mtpi*, the MTPI assumed q-axis current command value Iq_mtpi*, and the electrical angle estimated angular velocity ωe in accordance with the PMSM voltage equations shown in equations (13.1) and (13.2). Note that "p" in equations (13.1) and (13.2) is a differential operator.

Note that equations (13.1) and (13.2) take into account the voltage drops "p·Ld·Id" and "p·Lq·Iq" (p-term voltages) across the inductance due to current changes caused by torque control. "Ld" represents the d-axis inductance of motor M, and "Lq" represents the q-axis inductance of motor M.

Here, the p-term voltage is expressed using the time derivative of the current change. However, if the change in the detected current were used as the derivative value as is, the MTPI assumed d-axis voltage Vd_mtpi* and the MTPI assumed q-axis voltage Vq_mtpi* would be sensitive to current noise. Therefore, the derivative value is generated based on the current fundamental wave fluctuation, for example, as follows:

To explain the generation of the p-term voltage, first, the fluctuation components ΔIda and ΔIqa of the d-axis current Id and the q-axis current Iq are defined as in equations (14.1) and (14.2).

Here, when one periodic fluctuation occurs per mechanical angle rotation, "Ida" and "φd" in equation (14.1) indicate the fluctuation amplitude and initial phase of ΔIda, respectively, "Iqa" and "φq" in equation (14.2) indicate the fluctuation amplitude and initial phase of ΔIqa, respectively, and "θm" in equations (14.1) and (14.2) indicates the instantaneous value of the mechanical angle phase.

Therefore, the p-term voltage generated by the current fundamental wave fluctuation is expressed as in equations (15.1) and (15.2). That is, the phases of the d-axis current fluctuation and the q-axis current fluctuation are advanced by π/2, and the d-axis current fluctuation and the q-axis current fluctuation whose phases are advanced by π/2 are multiplied by the estimated mechanical angle angular velocity ωm, thereby generating a differential value (p-term voltage).

The MTPI voltage amplitude calculator 14b1-3 calculates the MTPI assumed output voltage Va_mtpi* based on the MTPI assumed d-axis voltage Vd_mtpi* and the MTPI assumed q-axis voltage Vq_mtpi* according to equation (16).

The average output voltage generator 14b1-4 outputs an average output voltage command value Va0* adjusted so that the average values of the d-axis current Id and the q-axis current Iq, which fluctuate with each rotation of the rotor of the motor M, trace an MTPI curve (maximum torque/current control curve). For example, the average output voltage generator 14b1-4 calculates the d-axis current Idt on the MTPI curve from the current q-axis current Iq, and adjusts the average output voltage command value Va0* using PI control or the like so that there is no error between the calculated d-axis current Idt and the current d-axis current Id. The average output voltage generator 14b1-4 calculates the average output voltage command value Va0* according to, for example, equations (17.1) and (17.2). Furthermore, when the average output voltage command value Va0* exceeds the output voltage limit value Vdq_limit, the average output voltage generator 14b1-4 limits the average output voltage command value Va0* to the output voltage limit value Vdq_limit in accordance with equation (18). Limiting the average output voltage command value Va0* to the output voltage limit value Vdq_limit results in flux-weakening control. "Ψa" indicates the flux linkage of the motor M.

The MTPI voltage fluctuation component extractor 14b1-5 calculates the fluctuation amplitude |ΔVa_mtpi| and instantaneous value ΔVa_mtpi of the MTPI assumed output voltage Va_mtpi*, for example, as follows:

The MTPI voltage fluctuation component extractor 14b1-5 first separates the fundamental component of the MTPI assumed output voltage Va_mtpi* into a Fourier coefficient Va_mtpi_sin, which is a sine component, and a cosine component Va_mtpi_cos, according to equations (19.1) and (19.2). The MTPI voltage fluctuation component extractor 14b1-5 calculates the Fourier coefficient of the fundamental component of the MTPI assumed output voltage Va_mtpi* for each mechanical angle period, thereby extracting the fundamental component of the MTPI assumed output voltage Va_mtpi* from which harmonic components have been removed.

Next, the MTPI voltage fluctuation component extractor 14b1-5 calculates the amplitude |ΔVa_mtpi| of the fundamental component of the MTPI assumed output voltage Va_mtpi* according to equation (20) based on the Fourier coefficients Va_mtpi_sin and Va_mtpi_cos calculated according to equations (19.1) and (19.2). Note that because the Fourier coefficients Va_mtpi_sin and Va_mtpi_cos are values that are updated every mechanical angle period, the amplitude |ΔVa_mtpi| is also updated every mechanical angle period.

Then, the MTPI voltage fluctuation component extractor 14b1-5 calculates the instantaneous value ΔVa_mtpi of the fundamental wave component of the MTPI assumed output voltage Va_mtpi* according to equation (21).

The adder 14b1-6 calculates the MTPI assumed output voltage fluctuation peak value Va_mtpi_peak by adding the average output voltage command value Va0* and the amplitude |ΔVa_mtpi| of the fundamental wave component of the MTPI assumed output voltage Va_mtpi* according to equation (22).

The MTPI voltage amplitude limit processor 14b1-7 generates a variable output voltage limit command value ΔVa_limit_mtpi adjusted so that the MTPI expected output voltage fluctuation peak value Va_mtpi_peak, which is the sum result of the adder 14b1-6, is equal to or less than the output voltage limit value Vdq_limit, and outputs the generated variable output voltage limit command value ΔVa_limit_mtpi.

For example, the MTPI voltage amplitude limit processor 14b1-7 calculates the amplitude ratio scale of the output voltage fluctuation component by comparing the average output voltage command value Va0*, the MTPI assumed output voltage fluctuation peak value Va_mtpi_peak, and the output voltage limit value Vdq_limit, and generates the variable output voltage limit command value ΔVa_limit_mtpi by multiplying the MTPI assumed output voltage fluctuation component ΔVa_mtpi by the amplitude ratio scale. In this way, it is possible to generate a variable output voltage limit command value ΔVa_limit_mtpi that is in phase with the MTPI assumed output voltage fluctuation component ΔVa_mtpi.

For example, the MTPI voltage amplitude limit processor 14b1-7 calculates the amplitude ratio scale of the output voltage fluctuation component according to equations (23.1) to (23.3), and generates the fluctuation output voltage limit command value ΔVa_limit_mtpi based on the calculated amplitude ratio scale according to equation (23.4).

The adder 14b1-8 calculates the output voltage limit command value Va* by adding the average output voltage command value Va0* and the variable output voltage limit command value ΔVa_limit_mtpi in accordance with the equation (24). The adder 14b1-8 outputs the calculated output voltage limit command value Va* to the induced voltage command value calculator 14b2 and the voltage command value calculator 14b6.

In FIG. 5, the induced voltage command value calculator 14b2 calculates the induced voltage command value Vo* based on the output voltage limit command value Va* in accordance with the motor model equations shown in equations (25.1) and (25.2) based on the current d-axis current Id, q-axis current Iq, and electrical angle estimated angular velocity ωe. Details of the calculation of the induced voltage command value Vo* are provided below.

The voltage equations (d-axis voltage Vd, q-axis voltage Vq) of the PMSM, the theoretical formula for the output voltage amplitude Va, and the theoretical formula for the induced voltage Vo of the motor M are shown in equations (25.1) to (27).

Furthermore, from equations (25.1) to (27), the equation relating the output voltage limit command value Va* and the induced voltage command value Vo* is given by equation (28). Therefore, induced voltage command value calculator 14b2 calculates induced voltage command value Vo* in accordance with equation (28) and outputs the calculated induced voltage command value Vo* to current command value calculator 14b3.

The current command value calculator 14b3 calculates the q-axis current command value Iq* and the d-axis current command value Id* based on the intersection of a constant torque curve, which is the current trajectory at which the total torque command value T* is constant, and a constant induced voltage ellipse, which is the current trajectory at which the induced voltage command value Vo* and the electrical angle estimated angular velocity ωe are constant (see FIG. 1B). The current command value calculator 14b3 outputs the calculated q-axis current command value Iq* and d-axis current command value Id* to the temporary voltage command value calculator 14b4.

The intersection of the constant torque curve and the constant induced voltage ellipse can be calculated using, for example, the motor torque equation shown in equation (29) and the induced voltage equation shown in equation (30).

By eliminating the d-axis current Id from equations (29) and (30), a quartic equation related to the q-axis current Iq can be obtained as shown in equation (31), where ΔL=Ld−Lq.

By applying, for example, Newton's method to the quartic equation shown in equation (31), it is possible to derive a solution corresponding to the q-axis current command value Iq* at the point where the constant torque curve, which is the current locus at which the total torque command value T* is constant, intersects with the constant induced voltage ellipse, which is the current locus at which the induced voltage Vo and the estimated electrical angle angular velocity ωe are constant (see Figure 1B).

After calculating the q-axis current command value Iq*, the current command value calculator 14b3 calculates the d-axis current command value Id* based on the q-axis current command value Iq* in accordance with equation (32), which is obtained by transforming the induced voltage equation shown in equation (30) into a d-axis current equation.

Whether to use a positive or negative sign for the √ in equation (32) can be determined by calculating the torque at the point where the constant induced voltage ellipse intersects with a line (hereinafter sometimes referred to as the "M-point boundary line") that is parallel to the Iq axis and passes through point M (-Ψa/Ld, 0), the center of the constant induced voltage ellipse, and comparing the M-point boundary torque T_M with the total torque command value T*.

The procedure for calculating the d-axis current command value Id* and the q-axis current command value Iq* is shown below. Figures 7A and 7B are diagrams used to explain an example of the operation of the current command value calculator according to this embodiment.

The current command value calculator 14b3 first calculates the d-axis current Id_M at point M according to equation (33).

Next, the current command value calculator 14b3 calculates the q-axis current Iq_M at the point where the boundary line at point M intersects with the constant induced voltage ellipse. The q-axis current Iq_M can be calculated by substituting the d-axis current Id_M at point M into equation (30), and is therefore calculated according to equation (34).

Therefore, the current command value calculator 14b3 calculates the M-point boundary torque T_M according to the equation (35).

The current command value calculator 14b3 determines the d-axis current command value Id* in accordance with equations (36.1) and (36.2) based on the magnitude relationship between the total torque command value T* and the torque on the boundary at point M. The equation (36.1) shows the d-axis current command value Id* (see FIG. 7A) when "the total torque command value T*≦the torque on the boundary at point M T_M", and the equation (36.2) shows the d-axis current command value Id* (see FIG. 7B) when "the total torque command value T*>the torque on the boundary at point M T_M".

The current command value calculator 14b3 outputs the d-axis current command value Id* and q-axis current command value Iq* calculated as described above to the temporary voltage command value calculator 14b4.

5 , the provisional voltage command value calculator 14b4 calculates a provisional d-axis voltage command value Vd_m and a provisional q-axis voltage command value Vq_m in a feedforward manner based on the electrical angle estimated angular velocity ωe, the d-axis current command value Id*, and the q-axis current command value Iq* in accordance with the motor model equations shown in Equations (37.1) and (37.2). The provisional voltage command value calculator 14b4 outputs the calculated provisional d-axis voltage command value Vd_m and provisional q-axis voltage command value Vq_m to the voltage vector angle calculator 14b5. Note that calculating the provisional voltage command values in a feedforward manner can prevent windup (saturation phenomenon) that occurs in an integrator such as in PI control due to input saturation.

Note that equations (37.1) and (37.2) take into account the voltage drops "p·Ld·Id" and "p·Lq·Iq" (p-term voltages) across the inductance due to current changes caused by torque control.

The voltage vector angle calculator 14b5 calculates a voltage vector angle δ based on the virtual d-axis voltage command value Vd_m and the virtual q-axis voltage command value Vq_m according to equation (38). The voltage vector angle calculator 14b5 outputs the calculated voltage vector angle δ to the voltage command value calculator 14b6. That is, as shown in FIG. 8 , the voltage vector angle calculator 14b5 calculates the angle between the q-axis and the output voltage vector having amplitude Va calculated using equation (1). This allows the voltage vector angle δ corresponding to the total torque command value T* to be calculated in a voltage saturation region where the output voltage amplitude is limited to or below the direct current (DC) voltage that the inverter can output. Therefore, vibration suppression control can be performed without tuning the voltage vector angle fluctuation, thereby improving the vibration suppression effect of the motor M without tuning the voltage vector angle fluctuation. FIG. 8 is a diagram illustrating an example of the operation of the voltage vector angle calculator according to this embodiment.

The voltage command value calculator 14b6 calculates the d-axis voltage command value Vd* and the q-axis voltage command value Vq* by converting the polar coordinates into the Cartesian coordinates based on the voltage vector angle δ and the output voltage limit command value Va* in accordance with equations (39.1) and (39.2).

<Operation of MTPI Voltage Limiting Processor>
FIG. 9 is a diagram illustrating an example of the operation of the MTPI voltage amplitude limiting processor according to this embodiment.

For example, as shown in case (a) in Figure 9, if the peak value Va_mtpi_peak of the MTPI assumed output voltage fluctuation component ΔVa_mtpi, which fluctuates around the average output voltage command value Va0*, is equal to or less than the output voltage limit value Vdq_limit, the condition of equation (23.2) is met, and the MTPI voltage amplitude limit processor 14b1-7 sets the amplitude ratio scale of the output voltage fluctuation component to "1." Then, the MTPI voltage amplitude limit processor 14b1-7 sets "scale = 1" in equation (23.4) and outputs the MTPI assumed output voltage fluctuation component ΔVa_mtpi as is as the variable output voltage limit command value ΔVa_limit_mtpi. As a result, the output voltage limit command value Va* matches the MTPI assumed output voltage fluctuation component ΔVa_mtpi.

For example, as shown in case (b) of Figure 9, if the peak value Va_mtpi_peak of the MTPI assumed output voltage fluctuation component ΔVa_mtpi, which fluctuates around the average output voltage command value Va0*, exceeds the output voltage limit value Vdq_limit, and the average output voltage command value Va0* does not exceed the output voltage limit value Vdq_limit, the condition of equation (23.3) is met, and therefore the MTPI voltage amplitude limit processor 14b1-7 sets the amplitude ratio scale of the output voltage fluctuation component to "(Vdq_limit-Va0*)/|ΔVa_mtpi|". The MTPI voltage amplitude limit processor 14b1-7 then outputs the variable output voltage limit command value ΔVa_limit_mtpi, which is calculated by using equation (23.4) with "scale = (Vdq_limit - Va0*) / |ΔVa_mtpi|". As a result, an output voltage limit command value Va* is generated that is in phase with the MTPI assumed output voltage fluctuation component ΔVa_mtpi and whose peak value due to the fluctuation amplitude is equal to or less than the output voltage limit value Vdq_limit.

For example, as shown in case (c) in Figure 9, if the average output voltage command value Va0* of the MTPI assumed output voltage fluctuation component ΔVa_mtpi is equal to or greater than the output voltage limit value Vdq_limit, the condition of equation (23.1) is met, and the MTPI voltage amplitude limit processor 14b1-7 sets the amplitude ratio scale of the output voltage fluctuation component to "0." The MTPI voltage amplitude limit processor 14b1-7 then sets "scale = 0" in equation (23.4) and outputs the variable output voltage limit command value ΔVa_limit_mtpi as "0." As a result, the output voltage limit command value Va* matches the output voltage limit value Vdq_limit.

By controlling the output voltage limit command value Va* in this manner, as shown in the example of the output voltage waveform shown in Figure 10, even immediately after the control region of motor M transitions from the normal control region to the voltage saturation region, the output voltage amplitude Va can be kept below the output voltage limit value Vdq_limit, and the output voltage amplitude Va can be matched between the normal control region and the voltage saturation region. This reduces switching shock when transitioning from the normal control region to the voltage saturation region. Furthermore, by including the voltage saturation region voltage command value generator 14b, the motor control device 100 can also handle cases in which the average output voltage command value Va0*, which is the center of fluctuation of the output voltage limit command value Va*, is limited by the output voltage limit value Vdq_limit in the voltage saturation region.

<Operation of normal control region voltage command value generator>
When the control switching determination unit 15 determines that the current control region of the motor M is the normal control region, the normal restriction region voltage command value generator 14a calculates the d-axis voltage command value Vd* and the q-axis voltage command value Vq* based on the total torque command value T*, the electrical angle estimated angular velocity ωe, the d-axis current Id, the q-axis current Iq, and the mechanical angle phase θm.

The normal limit region voltage command value generator 14a performs MTPI control (maximum torque/current control) to control torque, for example. With this method, the d-axis current command value Id* and the q-axis current command value Iq* are calculated based on the intersection of the MTPI curve and a constant torque curve, which is the current trajectory where the total torque command value T* is constant. Furthermore, the q-axis voltage command value Vq* is calculated by performing PI control on the difference between the q-axis current command value Iq* and the q-axis current Iq (q-axis current error). Furthermore, the d-axis voltage command value Vd* is calculated by performing PI control on the difference between the d-axis current command value Id* and the d-axis current Id (d-axis current error).

The d-axis current command value Id* and the q-axis current command value Iq* may be corrected for phase and amplitude errors that occur due to response delays when calculating the intersection of the constant torque curve and the MTPI curve, or due to interference between the d and q axes. The d-axis voltage command value Vd* and the q-axis voltage command value Vq* may be output as the final voltage command value after undergoing non-interference control. There are no limitations on the method for calculating the d-axis voltage command value Vd* and the q-axis voltage command value Vq*, and torque control methods other than MTPI control may also be used.

<Issues with rotor position estimation using steady-state models>
So far, we have described a method for calculating the control command values (d-axis voltage command value Vd* and q-axis voltage command value Vq*) for the motor M in the normal control region and the voltage saturation region. Below, we will describe a method for estimating the rotor position of the motor M.

As a method for estimating the rotor position from the d-axis current Id and the q-axis current Iq, the equation for calculating the axis error in a steady state using the extended induced voltage model shown in equation (40) has been widely used.

In equation (40), R is the resistance value of motor M, and Lq is the q-axis inductance of motor M. Furthermore, Vd and Vq are the d-axis voltage and q-axis voltage applied to motor M. Id and Iq are the d-axis current and q-axis current flowing through motor M. Note that the numerator in the arctangent function (tan ^-1 ) is the γ-axis induced voltage, and the denominator is the δ-axis induced voltage. Here, the γ-axis and δ-axis are estimated rotating coordinate systems for control purposes.

When estimating the rotor position, the axis error Δθ is calculated from equation (40), the axis error Δθ is used to estimate the electrical angle estimated angular velocity ωe, and the electrical angle phase (dq-axis phase) θe and mechanical angle phase θm, which represent the rotor position, are estimated using the electrical angle estimated angular velocity ωe. Furthermore, in calculating the axis error Δθ, the d-axis voltage command value Vd* and the q-axis voltage command value Vq* are used as Vd and Vq in equation (40), and the d-axis current Id and q-axis current Iq, which are detected values of the motor currents, are used as Id and Iq.

Here, the relationship between the above-mentioned torque control and motor current (d-axis current Id and q-axis current Iq) will be explained. When the load torque acting on motor M fluctuates periodically, such as when operating a single rotary compressor, unless the torque of motor M is adjusted, the rotational speed of motor M (electrical angle estimated angular velocity ωe) may also fluctuate periodically due to the periodic fluctuations in the load torque. In this case, there is a possibility that vibrations and noise may be generated, and stable control of motor M may become difficult. The above-mentioned torque control adjusts the torque of motor M in accordance with fluctuations in load torque in order to suppress such speed fluctuations.

In torque control, the d-axis current command value Id* and the q-axis current command value Iq* are adjusted to generate the desired torque, and the d-axis voltage command value Vd* and the q-axis voltage command value Vq* are generated based on the adjusted d-axis current command value Id* and q-axis current command value Iq*. In this case, it is expected that the d-axis current Id and the q-axis current Iq flowing through the motor M will also fluctuate periodically in accordance with the periodic fluctuations in the load torque.

The inventors have noticed that when torque control is performed using the above equation (40) in a situation where the d-axis current Id and the q-axis current Iq fluctuate periodically, oscillations occur near the peak points of the periodically fluctuating currents (hereinafter sometimes referred to as current peaks), which can result in increased power consumption. In particular, in the voltage saturation region where the output voltage amplitude Va of motor M is limited by the DC voltage, it has been found that significant oscillations occur near the current peaks during torque control that suppresses both voltage fluctuations and speed fluctuations (torque control performed by the voltage saturation region voltage command value generator 14b described above) (see Figure 12, etc.). This can lead to unstable control of motor M and increased power consumption.

Below, we will consider the reasons for current oscillation with reference to Figure 10. For example, Figure 10 shows how the output voltage amplitude Va periodically fluctuates in the normal control region. This periodic fluctuation in the output voltage amplitude Va is the result of controlling the torque of motor M in accordance with the periodic fluctuation in load torque. Note that in the normal control region, the output voltage amplitude Va is equal to or less than the output voltage limit value Vdq_limit. Therefore, there is no need to limit the output voltage amplitude Va. In this situation where the output voltage amplitude Va is not limited, for example, control is performed to periodically fluctuate the q-axis current command value Iq*, which primarily generates torque in motor M, out of the d-axis current command value Id* and the q-axis current command value Iq*. In this case, the amount of fluctuation in the d-axis current command value Id* is smaller than the q-axis current command value Iq*.

On the other hand, in the voltage saturation region, the output voltage amplitude Va is limited to an upper limit of the output voltage limit value Vdq_limit. The voltage saturation region voltage command value generator 14b adjusts the d-axis current command value Id* and the q-axis current command value Iq* so that the output voltage amplitude Va does not exceed the output voltage limit value Vdq_limit and so that speed fluctuations are suppressed. In this case, not only the amount of fluctuation in the q-axis current command value Iq* but also the amount of fluctuation in the d-axis current command value Id* increases. As a result, the d-axis current Id and the q-axis current Iq flowing through the motor M both fluctuate periodically by relatively large amounts. In this way, torque control in the voltage saturation region can be said to suppress both voltage fluctuations and speed fluctuations by varying the d-axis current Id.

The extended induced voltage model expressed by equation (40) is a steady-state model that assumes a steady state. In the steady state, the d-axis and q-axis currents (and the γδ-axis currents) are considered to be direct currents. Therefore, equation (40) does not take into account fluctuations in the d-axis current Id and the q-axis current Iq (i.e., changes over time in the d-axis current Id and the q-axis current Iq), and does not reflect the transient response of the current when estimating the position error Δθ. For this reason, it is thought that the fluctuation components of the d-axis current Id and the q-axis current Iq become larger, reducing the accuracy of estimating the position error Δθ. As a result, it is thought that position deviations occur during torque control, causing oscillations near the current peaks.

<Rotor position estimation using a transient model that takes current fluctuations into account>
The inventors focused on this point and constructed a transient model that takes into account fluctuations in the d-axis current Id and the q-axis current Iq. Equation (41) is the axis error calculation formula for the transient model. In equation (41), "p" is a differential operator, and Ld is the d-axis inductance of the motor M.

Equation (41) includes the inductance voltages "p·Ld·Id" and "p·Ld·Iq" (p-term voltages) that accompany current changes due to torque control. In other words, the transient model that takes into account current fluctuations shown in equation (41) takes into account the inductance voltage (p-term voltage), which is the differential value of the current. This makes it possible to reflect the effect of current fluctuations (effects due to transient changes in current) in torque control, etc.

For example, the extended induced voltage model, which is based on the steady state expressed in equation (40), does not take into account the inductance voltage that accompanies current changes, resulting in a model error. In particular, for the axis error Δθ to be zero, the accuracy of the model of the γ-axis induced voltage, which is the numerator in the arctangent function, is important. However, the numerator in the arctangent function of equation (40) does not include the inductance voltage (p Ld Id) that occurs due to current changes in the d-axis current Id (d-axis current fluctuations). For this reason, the model error in equation (40) becomes large, especially during torque control in the voltage saturation region where the d-axis current Id fluctuates greatly, and it is thought that this causes significant oscillation near the current peak.

In contrast, the model shown in equation (41) reflects the inductance voltage (p Ld Id) generated by changes in the d-axis current Id when calculating the axis error Δθ. This makes it possible to accurately calculate the axis error Δθ even when the d-axis current Id is varied during torque control in the voltage saturation region. Below, we will specifically explain the operation of the current fluctuation extractor 35, inductance voltage calculator 36, and axis error calculator 30 shown in Figure 2 as part of the process of calculating the axis error Δθ using the model shown in equation (41).

<Operation of the current fluctuation extractor>
The current fluctuation extractor 35 extracts current fluctuation components, which are periodic fluctuation components of the motor current. The current fluctuation components are periodic fluctuation components of the d-axis current Id and the q-axis current Iq that occur due to torque control. The current fluctuation extractor 35 extracts a d-axis current fluctuation component ΔId from the d-axis current Id and a q-axis current fluctuation component ΔIq from the q-axis current Iq.

In this embodiment, the current fluctuation extractor 35 extracts current fluctuation components by expanding the motor current (d-axis current Id and q-axis current Iq) into a Fourier series. Specifically, a calculation device such as a microcomputer that constitutes the motor control device 100 executes a process to calculate the Fourier coefficients of the Fourier series expansion for each of the d-axis current Id and the q-axis current Iq. This process is equivalent to, for example, a process that extracts periodic fluctuation components by Fourier transforming the d-axis current Id and the q-axis current Iq.

As will be described later, the current differential value, which is the differential of the current fluctuation component, is used to calculate the inductance voltage (p-term voltage). The current differential value is a parameter whose value is easily affected by noise, etc. In contrast, by using Fourier series expansion, it is possible to extract the current fluctuation component in which the effects of noise on the d-axis current Id and q-axis current Iq are suppressed. This stabilizes the value of the current differential value, making it possible to improve the accuracy of calculating the inductance voltage.

In addition, in this embodiment, the current fluctuation extractor 35 extracts two or more current fluctuation components with different periods. For example, when operating a single rotary compressor, the load torque oscillates with the same period as the mechanical angle phase θm of the motor M. In this case, when torque control is performed, higher-order current fluctuation components are generated in the d-axis current Id and the q-axis current Iq, such as a first-order current fluctuation component that fluctuates with the same period as the mechanical angle phase θm, and a second-order current fluctuation component that fluctuates with twice the period of the mechanical angle phase θm. The current fluctuation extractor 35 extracts the first-order current fluctuation component and higher-order current fluctuation component from each of the d-axis current Id and the q-axis current Iq. Using these current fluctuation components makes it possible to improve the accuracy of calculating the inductance voltage.

In this embodiment, the current fluctuation extractor 35 extracts the first-order and second-order current fluctuation components of the motor currents (d-axis current Id and q-axis current Iq) by expanding the motor currents (d-axis current Id and q-axis current Iq) into a Fourier series. For example, the more types of current fluctuation components used to estimate the rotor position, the more accurate the rotor position estimation will be, but the longer the processing time will be. Therefore, by using only the first-order and second-order current fluctuation components, it is possible to achieve both high accuracy in position estimation and reduced processing time.

In the following, the order of the current fluctuation component will be referred to as n (n = 1 or 2). The nth-order d-axis current fluctuation component will be referred to as ΔIdn, and the nth-order q-axis current fluctuation component will be referred to as ΔIqn.

First, a method for calculating the nth-order d-axis current fluctuation component ΔIdn will be described. The current fluctuation extractor 35 separates the nth-order d-axis current fluctuation component ΔIdn into a sine component and a cosine component. Specifically, the current fluctuation extractor 35 calculates the sine component Idn_sin of the nth-order d-axis current fluctuation component ΔIdn according to equation (42.1), and calculates the cosine component Idn_cos of the nth-order d-axis current fluctuation component ΔIdn according to equation (42.2).

Equations (42.1) and (42.2) represent the nth-order Fourier coefficients. The nth-order mechanical angle phase θmn is the first-order mechanical angle phase θm1 when n = 1, and is the second-order mechanical angle phase θm2 when n = 2. The nth-order mechanical angle phase θmn (θm1 or θm2) is an instantaneous value calculated by the position estimator 32 described with reference to Figure 2. The sine component Idn_sin and cosine component Idn_cos of the nth-order d-axis current fluctuation component ΔIdn are calculated for each mechanical angle cycle. This makes it possible to remove noise and other harmonic components.

The current fluctuation extractor 35 calculates the amplitude |ΔIdn| of the nth-order d-axis current fluctuation component ΔIdn according to equation (43.1) using the sine component Idn_sin and cosine component Idn_cos of the nth-order d-axis current fluctuation component ΔIdn, and calculates the reference phase φdn of the nth-order d-axis current fluctuation component ΔIdn according to equation (43.2).

The n-th order d-axis current fluctuation component ΔIdn is expressed as in equation (44) using the amplitude |ΔIdn| of equation (43.1) and the reference phase φdn of equation (43.2).

The nth-order mechanical angle phase θmn written in equation (44) is a variable that changes with the rotation of the motor M. Therefore, the waveform of the nth-order d-axis current fluctuation component ΔIdn is a sine waveform that oscillates with an amplitude |ΔIdn|, with the reference phase φdn as the initial phase. In this way, the process of extracting the nth-order d-axis current fluctuation component ΔIdn is a process of calculating parameters (amplitude |ΔIdn| and reference phase φdn) that can represent that waveform.

Next, a method for calculating the nth-order q-axis current fluctuation component ΔIqn will be described. The method for calculating the nth-order q-axis current fluctuation component ΔIqn is basically the same as the method for calculating the nth-order d-axis current fluctuation component ΔIdn. Specifically, the current fluctuation extractor 35 calculates the sine component Iqn_sin of the nth-order q-axis current fluctuation component ΔIqn according to equation (45.1), and calculates the cosine component Iqn_cos of the nth-order q-axis current fluctuation component ΔIqn according to equation (45.2).

In addition, the current fluctuation extractor 35 calculates the amplitude |ΔIqn| of the nth-order q-axis current fluctuation component ΔIqn according to equation (46.1) using the sine component Iqn_sin and cosine component Iqn_cos of the nth-order q-axis current fluctuation component ΔIqn, and calculates the reference phase φqn of the nth-order q-axis current fluctuation component ΔIqn according to equation (46.2).

The n-th order q-axis current fluctuation component ΔIqn is expressed as in equation (47) using the amplitude |ΔIqn| of equation (46.1) and the reference phase φqn of equation (46.2).

As shown in equation (47), the waveform of the nth-order q-axis current fluctuation component ΔIqn is a sine waveform that oscillates with an amplitude |ΔIqn|, with the reference phase φqn as the initial phase. In this way, the process of extracting the nth-order q-axis current fluctuation component ΔIqn is a process of calculating parameters (amplitude |ΔIqn| and reference phase φqn) that can represent that waveform.

The current fluctuation components calculated by the current fluctuation extractor 35 through Fourier series expansion are shown in equations (48.1) to (48.4). Equation (48.1) is the first-order d-axis current fluctuation component ΔId1, and equation (48.2) is the second-order d-axis current fluctuation component ΔId2. Equation (48.3) is the first-order q-axis current fluctuation component ΔIq1, and equation (48.4) is the second-order q-axis current fluctuation component ΔIq2.

The current fluctuation extractor 35 calculates the amplitudes (|ΔId1|, |ΔId2|, |ΔIq1|, |ΔIq2|) and reference phases (φd1, φd2, φq1, φq2) in each of equations (48.1) to (48.4) as the current fluctuation components. These parameters are output to the inductance voltage calculator 36.

<Operation of the inductance voltage calculator>
The inductance voltage calculator 36 calculates the inductance voltage, which is an induced voltage generated by the current fluctuation component, based on the inductance of the motor M and the differential value of the current fluctuation component. Here, the inductance voltage is the p-term voltage (p·Ld·Id and p·Ld·Iq) shown in equation (41). Therefore, the d-axis inductance Ld is used as the inductance of the motor M.

Furthermore, the differential operator "p" acts only on the fluctuation component of the parameter being calculated (the differential value for DC components is 0). Therefore, the current fluctuation components (ΔId1, ΔId2, ΔIq1, ΔIq2) calculated by the current fluctuation extractor 35 can be used as the d-axis current Id and q-axis current Iq in the p-term voltage.

Here, the current component of the d-axis current Id that is correlated with the first-order mechanical angle phase θm1 is referred to as the current correlation value Id1, and the current component that is correlated with the second-order mechanical angle phase θm2 is referred to as the current correlation value Id2. The current correlation value Id1 is the first-order d-axis current fluctuation component ΔId1, and the current correlation value Id2 is the second-order d-axis current fluctuation component ΔId2. Similarly, the current component of the q-axis current Iq that is correlated with the first-order mechanical angle phase θm1 is referred to as the current correlation value Iq1, and the current component that is correlated with the second-order mechanical angle phase θm2 is referred to as the current correlation value Iq2. The current correlation value Iq1 is the first-order q-axis current fluctuation component ΔIq1, and the current correlation value Iq2 is the second-order q-axis current fluctuation component ΔIq2.

For example, for the d-axis current Id, the inductance voltage due to current fluctuations with a period similar to that of the primary mechanical angle phase θm1 is calculated using the current correlation value Id1 according to equations (49.1) to (49.4).

In this case, as shown in equation (49.1), the current correlation value Id1 is used as the d-axis current Id of the p-term voltage, and the current correlation value Id1 is time-differentiated. In equation (49.2), the time derivative of the current correlation value Id1 is converted into a derivative with respect to the first-order mechanical angle phase θm1. As a result, a term (dθm1/dt) that time-differentiates the first-order mechanical angle phase θ1 is generated, and as shown in equation (49.3), this term is expressed as the estimated mechanical angle angular velocity ωm. Furthermore, in equation (49.3), the current correlation value Id1 is replaced with the first-order d-axis current fluctuation component ΔId1.

By applying the above equation (48.1) to equation (49.3) and differentiating the first-order d-axis current fluctuation component ΔId1 with respect to the first-order mechanical angle phase θm1, equation (49.4) is obtained. In this way, the inductance voltage (p·Ld·Id1) corresponding to the first-order component of the d-axis current Id can be calculated using the estimated mechanical angle angular velocity ωm, the d-axis inductance Ld, the amplitude |ΔId1|, and the reference phase φd1. Note that the first-order mechanical angle phase θm1 described in equation (49.4) is a variable that changes with the rotation of motor M. Therefore, by inputting the instantaneous value of the first-order mechanical angle phase θm1 into equation (49.4), the value of the inductance voltage (p·Ld·Id1) at any timing can be calculated.

The inductance voltage (p·Ld·Iq1) corresponding to the primary component of the q-axis current Iq can also be calculated in the same manner as above. p·Ld·Iq1 is shown in equation (50).

Similarly, the inductance voltage (p·Ld·Id2) corresponding to the second-order component of the d-axis current Id and the inductance voltage (p·Ld·Iq2) corresponding to the second-order component of the q-axis current Iq can be calculated. Equation (51) shows p·Ld·Id2, and equation (52) shows p·Ld·Iq2.

In the process of deriving the second-order component of the inductance voltage, the time derivative (dθm2/dt) of the second-order mechanical angle phase θm2 is obtained. The value of dθm2/dt is twice the time derivative of the first-order mechanical angle phase θm1, or 2ωm, as shown in equation (53).

In this way, the inductance voltage calculator 36 calculates the inductance voltage for each of two or more current fluctuation components with different frequencies based on the differential value of the current fluctuation component. In this embodiment, for example, a primary inductance voltage and a secondary inductance voltage are calculated for the d-axis current Id (or the q-axis current Iq), and both are induced voltages that occur in response to current fluctuations in the d-axis current Id (or the q-axis current Iq).

The induced voltage generated in response to current fluctuations in the d-axis current Id (or the q-axis current Iq) can be calculated as the sum of the primary inductance voltage and the secondary inductance voltage. Equation (54) represents the inductance voltage p Ld Id due to the fluctuation component of the d-axis current Id, and Equation (55) represents the inductance voltage p Ld Iq due to the fluctuation component of the q-axis current Iq.

In this way, the inductance voltage calculator 36 calculates the combined inductance voltage (p·Ld·Id and p·Ld·Iq) by adding up the inductance voltages for two or more current fluctuation components with different frequencies. This makes it possible to reflect induced voltages caused by not only primary current fluctuations but also higher-order current fluctuations, improving the accuracy of inductance voltage calculations. This inductance voltage is used to calculate the axis error Δθ.

For example, the inductance voltages p, Ld, and Id generated by d-axis current fluctuations are calculated by inputting the instantaneous values of the first and second mechanical angle phases (θm1 and θm2) into equations (49.4) and (51) and adding these values according to equation (54). Similarly, the inductance voltages p, Ld, and Iq generated by q-axis current fluctuations are calculated by inputting the instantaneous values of the first and second mechanical angle phases (θm1 and θm2) into equations (50) and (52) and adding these values according to equation (54). Each calculated inductance voltage is output to the axis error calculator 30.

<Operation of the axis error calculator>
The axis error calculator 30 calculates the axis error Δθ in accordance with the axis error calculation formula in the transient model shown in formula (41). This makes it possible to accurately calculate the axis error Δθ by taking into account the inductance voltage (p-term voltage) that accompanies periodic current fluctuations in the d-axis current Id and q-axis current Iq that occur during torque control, etc.

For example, for terms other than the inductance voltage (p-term voltage), the d-axis voltage command value Vd* and the q-axis voltage command value Vq* are used as Vd and Vq, and the d-axis current Id and q-axis current Iq, which are detected values of the motor current, are used as Id and Iq. Furthermore, the inductance voltage (p-term voltage) uses the inductance voltage p·Ld·Id generated by d-axis current fluctuations and the inductance voltage p·Ld·Iq generated by q-axis current fluctuations output from the inductance voltage calculator 36.

As a result, the axis error Δθ reflects the effect of the induced voltage (inductance voltage) caused by the current fluctuation component. The axis error Δθ is also used in the estimation process of the rotor position (mechanical angle phase θm and electrical angle phase θe). In this way, the motor control device 100 estimates the rotor position of the motor M based on the inductance voltage. This makes it possible to estimate the rotor position by taking into account the inductance voltage, which is the induced voltage caused by the current fluctuation component, and improves the accuracy of the rotor position estimation.

In addition, in this embodiment, the total inductance voltage shown in the above equations (54) and (55) is used as the inductance voltage. In other words, the rotor position is estimated based on the total inductance voltage. This makes it possible to reflect the influence of induced voltage due to frequency components other than the frequency of the load torque, for example, and further improves the accuracy of rotor position estimation.

<Operation results for steady-state and transient models>
In the following, motor control using the steady-state model shown in equation (40) will be compared as a comparative example with the operation results of motor control using the transient model according to this embodiment shown in equation (41).

Figure 11 is a graph showing measurement data during motor control using a steady-state model (a comparative example) and the transient model according to this embodiment. The horizontal axis of each graph shown in Figure 11 represents time t [msec]. This graph shows data obtained when switching from motor control using the steady-state model to motor control using the transient model during control of the motor M that operates a single rotary compressor.

The timing for switching from the steady-state model to the transient model was t = 8000 msec. Each piece of data was measured during torque control (torque control performed by the voltage saturation region voltage command value generator 14b) that suppresses voltage fluctuations in the voltage saturation region. This measurement was also performed under an overload condition where a load above the rated capacity was applied to motor M in order to clearly demonstrate oscillations that occur near current peaks when the motor is controlled using the steady-state model. Even in normal operating conditions that are not overloaded, oscillations near current peaks occur when the steady-state model is used.

The graph on the top left of Figure 11 shows the mechanical angular velocity, plotting the estimated mechanical angular velocity ωm and mechanical angular velocity command value ωm* of motor M. The vertical axis of the graph represents rotational speed [rps (rotations per second)]. The mechanical angular velocity command value ωm* during measurement was unchanged before and after the switching timing, remaining at 35 rps.

The graph on the left side of the middle row in Figure 11 is a graph for the d-axis current and q-axis current, plotting the current command values (d-axis current command value Id* and q-axis current command value Iq*) and the detected current values (d-axis current Id and q-axis current Iq). The vertical axis of the graph represents the current value [A].

The graph on the bottom left of Figure 11 is a graph of the d-axis voltage and q-axis voltage, plotting the detected voltage values (d-axis voltage Vd and q-axis voltage Vq). The vertical axis of the graph represents the voltage value [V].

The graph on the upper right of Fig. 11 is a graph of the output voltage amplitude Va and the DC voltage Vdc. The vertical axis of the graph represents voltage [V]. Data in the dq dimension is used for the output voltage amplitude Va and the DC voltage Vdc. The output voltage amplitude Va in the dq dimension is expressed by the following equation (56) using the d-axis voltage Vd and the q-axis voltage Vq.
Furthermore, if the DC voltage Vdc in the dq dimension is Vdc_2φ and the DC voltage on the stationary coordinate system (uvw dimension) is Vdc_3φ, the relationship between Vdc_2φ and Vdc_3φ is expressed by the following equation (57).

The graph in the middle right of Figure 11 shows the axis error Δθ. The vertical axis of the graph represents angle [deg]. In the first half of the graph (t<8000 msec), the axis error Δθ was calculated according to equation (40). In the second half of the graph (t≧8000 msec), the axis error Δθ was calculated according to equation (41). Here, the axis error Δθ data is plotted after smoothing the values calculated from each equation using an LPF (low-pass filter) with a time constant of 1 msec.

The graph on the lower right side of Fig. 11 shows the power consumption Pinv of the inverter (IPM 25). The vertical axis of the graph represents power [W]. The power consumption Pinv is expressed by the following equation (58) using the d-axis voltage Vd, the q-axis voltage Vq, the d-axis current Id, and the q-axis current Iq.
Here, the data of Pinv is plotted by smoothing the calculated value from equation (58) using an LPF with a time constant of 1000 msec.

Figure 12 is a graph showing measurement data during motor control using a steady-state model, presented as a comparative example. Figure 12 shows measurement data from the period shown in Figure 11 prior to the switching timing t = 8000 msec (0 msec ≤ t ≤ 100 msec). The layout of each graph is the same as in Figure 11.

As mentioned above, the load torque of a single rotary compressor fluctuates periodically with a mechanical angle cycle. In this measurement, torque control was performed to suppress fluctuations in rotation speed caused by such fluctuations in load torque. This makes it possible to suppress the amplitude (fluctuation range) of the estimated mechanical angle angular velocity ωm compared to when torque control is not performed.

For example, in the graph in the upper left corner of the top row of Figure 12, the estimated mechanical angle angular velocity ωm fluctuates periodically. The period of this fluctuation is the same as the mechanical angle period, and it is thought that the periodic fluctuation of the estimated mechanical angle angular velocity ωm occurs due to the periodic fluctuation of the load torque of the single rotary compressor. In this measurement, by performing torque control, the fluctuation range of the estimated mechanical angle angular velocity ωm, excluding harmonic components, is suppressed to approximately ±4 rps relative to the mechanical angular velocity command value ωm*. Note that if torque control is not performed, the fluctuation range of the estimated mechanical angle angular velocity ωm would be larger.

In torque control, the d-axis current command value Id* and the q-axis current command value Iq* are adjusted to suppress fluctuations in the rotational speed (mechanical angle estimated angular velocity ωm) of the motor M. As a result, as shown in the graph in the middle left of Figure 12, both the d-axis current command value Id* and the q-axis current command value Iq* (dotted line graph) fluctuate with a mechanical angle period.

This measurement was also performed in the voltage saturation region. In this case, the voltage saturation region voltage command value generator 14b described above operates, adjusting the d-axis current command value Id* and the q-axis current command value Iq* so that the output voltage amplitude Va is equal to or less than the DC voltage Vdc. For example, under the conditions of this measurement, as shown in the graph, the d-axis current command value Id* was adjusted to fluctuate over a range similar to that of the q-axis current command Iq* that generates torque in the motor M. Accordingly, as shown in the graph on the bottom left of Figure 12, the d-axis voltage Vd also fluctuated relatively greatly.

In this way, torque control suppresses fluctuations in the output voltage amplitude Va by varying the d-axis voltage Vd (d-axis current command value Id*). When the axis error Δθ is estimated using the steady-state model shown in equation (40), oscillations are observed near the current peak. For example, the graph on the left side of the middle row of Figure 12 shows that the detected q-axis current Iq oscillates near the current peak compared to the q-axis current command value Iq*. Similarly, oscillations also occur in the d-axis current Id, and a deviation from the waveform of the d-axis current command value Id* is also observed.

Figure 13 is a graph showing measurement data during motor control using the transient model according to this embodiment. Figure 13 shows measurement data from the data shown in Figure 11 for a period after the switching timing t = 8000 msec (15000 msec ≤ t ≤ 15100 msec). The layout of each graph is the same as in Figure 11.

As shown in the graph in the middle left of Figure 13, when motor control is performed using the transient model, oscillations near the current peak of the q-axis current Iq are reduced compared to when the steady-state model is used (graph in the middle left of Figure 12). Furthermore, waveform distortion associated with oscillations in the d-axis current is eliminated, reducing the deviation of the waveform from the d-axis current command value Id*.

Furthermore, as shown in the graph on the lower left of Figure 13, the reduction in oscillation of the q-axis current Iq near the current peak also suppresses small vibrations in the q-axis voltage Vq near the current peak. Furthermore, the d-axis voltage Vd fluctuates in the same way as when the steady-state model is used, indicating that torque control in the voltage saturation region is maintained appropriately.

The reason why oscillations near the current peak are suppressed in this way is thought to be because the use of the transient model of equation (41) improves the accuracy of estimating the position error Δθ. For example, looking at the graph in the middle right of Figure 13, we can see that the position error Δθ calculated using the transient model has a smaller range of change in value compared to when the steady-state model is used (graph in the middle right of Figure 12). For example, the depth of the valley-shaped peaks that occur during the mechanical angle cycle and the range of fluctuations in the higher-frequency components than the mechanical angle cycle are both suppressed. In this way, using the transient model makes it possible to stably estimate the position error Δθ.

Furthermore, as shown in the graph on the top left of Figure 13, when the transient model is used, the periodic fluctuations in the estimated mechanical angle speed ωm are suppressed to a level equivalent to that when the steady-state model is used (graph on the top left of Figure 12). It can also be seen that fluctuation components with frequencies higher than the mechanical angle period are suppressed, enabling stable control of the rotation speed.

Furthermore, by suppressing oscillations near the current peak, it is possible to reduce power consumption. For example, looking at the graph on the bottom right of Figure 11, power consumption Pinv decreases around the switching timing, and eventually stabilizes at a certain level. In this way, the behavior of power consumption Pinv also shows that motor M is being controlled stably.

As described above, in the motor control device 100 according to this embodiment, the rotor position of the motor M is estimated based on the current fluctuation components (d-axis current fluctuation component ΔId and q-axis current fluctuation component ΔIq), which are periodic fluctuation components contained in the motor current (d-axis current Id and q-axis current Iq). This makes it possible to estimate the rotor position with high accuracy even when the motor current (d-axis current Id and q-axis current Iq) fluctuates periodically.

In this embodiment, the axis error calculation formula for the transient model shown in Equation (41) is used to estimate the axis error Δθ. Equation (41) is a formula for calculating the axis error Δθ by taking into account the inductance voltage generated by the periodic current fluctuation component. By introducing the inductance voltage, when the motor current fluctuates periodically, it is possible to reduce the model error of the extended induced voltage model constructed as a steady-state model shown in Equation (40). This makes it possible to improve the estimation accuracy of the rotor position.

This action makes it possible to suppress oscillations near current peaks, for example, under torque control where the motor current fluctuates periodically. This makes it possible to reduce the power consumption of the motor M. Furthermore, suppressing oscillations stabilizes the estimated value of the axis error Δθ, making it possible to stably estimate the rotor position. As a result, stable motor control can be achieved.

In addition, in this embodiment, focusing on the periodicity of current fluctuations under torque control, the motor current is expanded into a Fourier series (calculating the Fourier coefficients of the motor current) to extract the current fluctuation component. By differentiating this current fluctuation component, the current differential value, which becomes the inductance voltage (p-term voltage), is calculated. For example, in a method that generates the current differential value from the current change detected every control cycle without using Fourier coefficients, the current differential value may become unstable depending on current noise, which could cause control instability. In contrast, in this embodiment, the use of Fourier coefficients makes it possible to stably calculate the current differential value, making it possible to appropriately suppress model errors corresponding to the inductance voltage.

In addition, in this embodiment, in addition to the first-order current fluctuation component that fluctuates with the same period as the load torque, a second-order current fluctuation component that fluctuates with twice the period of the load torque is extracted as the current fluctuation component. This makes it possible to accurately estimate the position error Δθ by taking into account not only the first-order inductance voltage but also the second-order inductance voltage. Furthermore, by limiting the order of the extracted current fluctuation component to two, it is possible to improve the estimation accuracy of the position error Δθ without unnecessarily increasing the calculation load.

<Other embodiments>
The present invention is not limited to the above-described embodiment, and various other embodiments can be realized.

<Switching the axis error calculation method>
The following describes a configuration for switching the method for calculating the axis error Δθ.

In the above embodiment, a method for calculating the axis error Δθ from the axis error calculation formula for the transient model shown in equation (41) was described, regardless of whether the motor control region is the normal control region or the voltage saturation region. This is not limiting, and the method for calculating the axis error Δθ may be switched between the normal control region and the voltage saturation region.

For example, in the motor control device 100 shown in FIG. 1, the control switching determination unit 15 determines whether the current control region is the normal control region or the voltage saturation region. This determination result can be used to switch the formula used to calculate the axis error Δθ in the axis error calculator 30. In this configuration, if it is determined that the control region is the normal control region, the axis error Δθ is calculated using equation (40). In the normal control region, unlike the voltage saturation region, there is no need to limit the output voltage amplitude. Therefore, for example, the amount of fluctuation in the d-axis current command value Id* is small, and the periodic fluctuation component of the d-axis current Id is also small. For this reason, even when equation (40) is used, the axis error Δθ can be estimated with sufficient accuracy.

On the other hand, if it is determined that the motor is in the voltage saturation region, the axis error Δθ is calculated using equation (41). That is, if the motor's control region is in the voltage saturation region, the rotor position may be estimated based on the current fluctuation component. For example, in the voltage saturation region, the operation of the voltage saturation region voltage command value generator 14b described above increases the fluctuations in the d-axis current command value Id* as well as the q-axis current command value Iq, and the fluctuations in the d-axis current Id become large along with the q-axis voltage Iq. As a result, the effect of the inductance voltage associated with periodic current fluctuations cannot be ignored. In such cases, for example, using the transient model of equation (41) allows for more accurate estimation of the rotor position than using the steady-state model of equation (40).

In the above-described embodiment, torque control is performed to suppress fluctuations in rotational speed due to the load torque applied to the motor. In the motor control device 100 shown in FIG. 1, torque control is performed when the torque fluctuation command value ΔT output from the correction torque generator 34 is not 0 (ΔT ≠ 0). Note that when the torque fluctuation command value ΔT = 0, speed control is performed to control the rotational speed so as to achieve the mechanical angular velocity command value ωm*.

For example, if fluctuations in rotation speed are acceptable (e.g., if fluctuations in rotation speed are sufficiently small during high-speed rotation), torque control is not necessarily required. In this case, speed control is performed, for example, with a fluctuating torque command value ΔT = 0. In this way, the motor control device may be configured to be able to switch torque control ON and OFF. In such a configuration, when torque control is set to OFF, the axis error Δθ is calculated using equation (40). When torque control is not performed, the amount of fluctuation in the q-axis current command value Iq* and the d-axis current command value Id* is smaller than when torque control is performed, for example. Therefore, even when equation (40) is used, the axis error Δθ can be estimated with sufficient accuracy.

On the other hand, when torque control is set to ON, the axis error Δθ is calculated using equation (41). That is, when torque control is performed on the motor, the rotor position may be estimated based on the current fluctuation component. For example, during torque control, the q-axis current command value Iq* and the d-axis current command value Id* are also periodically varied in accordance with the periodic fluctuations in the load torque, resulting in periodic fluctuations in the q-axis voltage Iq and the d-axis current Id. In other words, periodic fluctuation components are more likely to occur in the motor current. In such cases, using the transient model of equation (41), for example, allows for more accurate estimation of the rotor position than using the steady-state model of equation (40).

In the above embodiment, the operation of a single rotary compressor has been mainly described as an example. In this case, the load torque fluctuates with a period similar to the mechanical angle period, so a first-order current fluctuation component that fluctuates with a period similar to the mechanical angle period and a second-order current fluctuation component with a period twice as long are used for motor control. There are no limitations on the type of compressor to which the present invention can be applied.

For example, a twin rotary compressor or the like may be used. In this case, the load torque will fluctuate at a period twice the mechanical angle period. Therefore, the current fluctuation components may be extracted as second-order current fluctuation components that fluctuate at a period twice the mechanical angle period, or fourth-order current fluctuation components that fluctuate at a period four times the mechanical angle period. In this way, by extracting current fluctuation components according to the order of torque control performed by the motor control device and using their current derivative values, it is possible to reliably improve the estimation accuracy of the axis error Δθ.

The above describes a method of calculating the inductance voltage from the current fluctuation component of the motor current and then using the inductance voltage to calculate the axis error Δθ from equation (41). For example, a method of adjusting the axis error Δθ using the current fluctuation component without calculating the inductance voltage may also be used. For example, an adjustment amount corresponding to the amplitude and phase of the current fluctuation component may be determined in advance through experiments or the like and used for motor control. In this case, the amplitude and phase of the current fluctuation component are calculated, and the axis error Δθ is successively adjusted using an adjustment amount according to the calculation result. Furthermore, for example, an adjustment amount according to the differential value of the current fluctuation component may also be used. Other methods for estimating the rotor position based on the current fluctuation component are not limited to these.

It is also possible to combine at least two of the features of the present technology described above. In other words, the various features described in each embodiment may be combined in any way, regardless of the embodiment. Furthermore, the various effects described above are merely examples and are not limiting, and other effects may also be achieved.

M...Motor 25...IPM
28... 3φ current calculator 29... u, v, w/dq converter 35... current fluctuation extractor 36... inductance voltage calculator 30... axis error calculator 31... PLL controller 32... position estimator 33... 1/Pn processor 100... motor control device

Claims (7)

a current detection unit that detects a motor current flowing through the motor;
a current fluctuation extraction unit that extracts a current fluctuation component that is a periodic fluctuation component of the motor current by expanding the motor current into a Fourier series ;
a position estimation unit that estimates a rotor position of the motor based on the current fluctuation component. 2. The motor control device according to claim 1,
The current fluctuation extraction unit extracts the current fluctuation component having the same period as the fluctuation period of the load torque acting on the motor and the current fluctuation component having a period that is an integer multiple of the fluctuation period of the load torque by expanding the motor current into a Fourier series.
Motor control device. 2. The motor control device according to claim 1, further comprising:
an inductance voltage calculation unit that calculates an inductance voltage, which is an induced voltage generated by the current fluctuation component, based on the inductance of the motor and a differential value of the current fluctuation component;
The motor control device, wherein the position estimation unit estimates the rotor position based on the inductance voltage. 4. The motor control device according to claim 3 ,
the current fluctuation extraction unit extracts two or more current fluctuation components having different periods;
the inductance voltage calculation unit calculates the inductance voltage for each of the two or more current fluctuation components based on a differential value of the current fluctuation component, and calculates a total inductance voltage by summing the inductance voltages for each of the two or more current fluctuation components;
The motor control device, wherein the position estimation unit estimates the rotor position based on the total inductance voltage. 5. The motor control device according to claim 4 ,
The motor control device, wherein the current fluctuation extraction unit extracts a first-order current fluctuation component and a second-order current fluctuation component of the motor current by expanding the motor current into a Fourier series. 2. The motor control device according to claim 1,
The motor control device, wherein the position estimator estimates the rotor position based on the current fluctuation component when performing torque control on the motor. 2. The motor control device according to claim 1,
The position estimation unit
When a control region of the motor is a voltage saturation region, the rotor position is estimated based on the current fluctuation component;
When the control region of the motor is a normal control region in which the output voltage amplitude of the motor is smaller than that of the voltage saturation region, the rotor position is estimated based on the motor current without extracting the current fluctuation component.
Motor control device.