Motor control device

Abstract

To provide a motor control device capable of estimating a rotor position using high-frequency current while suppressing motor noise.SOLUTION: In a motor control device 100d, a position estimator 32, a PLL controller 31, and an axis error calculator 30 estimate the position of a rotor that a motor M has based on a high-frequency current generated when a high-frequency voltage is applied to the motor M. A high-frequency calculator 81, an application method determiner 82, and a high-frequency voltage command value generator 43 apply high-frequency voltage to the motor while changing the frequency of the sideband waves generated by the high-frequency voltage applied to the motor.SELECTED DRAWING: Figure 13

Description

[Technical Field]

[0001] This disclosure relates to a motor control device. [Background technology]

[0002] One technique for estimating the rotor position in sensorless vector control for AC motors involves applying a high-frequency voltage that does not contribute to torque generation to the motor and estimating the rotor position using the high-frequency component (hereinafter sometimes referred to as "high-frequency current") contained in the detected current. In this technique, the rotor position is estimated based on a common-mode current vector that rotates in the same direction as the high-frequency magnetic flux vector generated in response to the application of the high-frequency voltage, and a mirror-mode current vector that rotates in the opposite direction to the high-frequency magnetic flux vector (Patent Document 1). Hereinafter, the common-mode current vector and the mirror-mode current vector will be collectively referred to as "common-mode mirror-mode current vector". [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2002-171799 [Overview of the project] [Problems that the invention aims to solve]

[0004] However, when high-frequency current flows through a motor, noise is generated.

[0005] Therefore, this disclosure proposes a technology that can estimate the rotor position using high-frequency current while suppressing motor noise. [Means for solving the problem]

[0006] The motor control device of this disclosure comprises an estimator and an application unit. The estimator estimates the position of the rotor of the motor based on the high-frequency current generated when a high-frequency voltage is applied to the motor. The application unit applies the high-frequency voltage to the motor. The application unit can also change the frequency of the sidebands generated by the high-frequency voltage. [Effects of the Invention]

[0007] According to this disclosure, the rotor position can be estimated using high-frequency current while suppressing motor noise. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows an example of the configuration of a motor control device according to Embodiment 1 of the present disclosure. [Figure 2] Figure 2 shows an example of the configuration of the noise reduction command generator in Embodiment 1 of this disclosure. [Figure 3] Figure 3 shows an example of the operation of the high-frequency voltage command value generator of Embodiment 1 of this disclosure. [Figure 4] Figure 4 shows an example of the configuration of the drive component removal filter in Embodiment 1 of this disclosure. [Figure 5] Figure 5 shows an example of the configuration of the axis error calculator in Embodiment 1 of this disclosure. [Figure 6] Figure 6 shows an example configuration of the in-phase mirror-phase current vector generator of Embodiment 1 of this disclosure. [Figure 7] Figure 7 shows an example of the operation of the count calculator, command generator, and high-frequency voltage command value generator of Embodiment 2 of this disclosure. [Figure 8] Figure 8 shows an example of the operation of the count calculator, command generator, and high-frequency voltage command value generator of Embodiment 3 of this disclosure. [Figure 9] Figure 9 shows an example of the operation of the high-frequency voltage command value generator of Embodiment 4 of this disclosure. [Figure 10] Figure 10 shows an example of the configuration of the axis error calculator in Embodiment 5 of this disclosure. [Figure 11]Figure 11 shows an example of the configuration of a motor control device according to Embodiment 6 of this disclosure. [Figure 12] Figure 12 shows an example of the configuration of a motor control device according to Embodiment 7 of this disclosure. [Figure 13] Figure 13 shows an example of the configuration of a motor control device according to Embodiment 8 of the present disclosure. [Figure 14] Figure 14 is a flowchart showing an example of the processing procedure in the high-frequency calculator of Embodiment 8 of this disclosure. [Figure 15] Figure 15 shows an example of the operation of the high-frequency calculator of Embodiment 8 of this disclosure. [Figure 16] Figure 16 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 17] Figure 17 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 18] Figure 18 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 19] Figure 19 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 20] Figure 20 is a diagram illustrating an example of the operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 21] Figure 21 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 22] Figure 22 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Figure 23] Figure 23 is a diagram illustrating an example of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure. [Modes for carrying out the invention]

[0009] The embodiments of this disclosure will be described below with reference to the drawings. In the following embodiments, the same parts will be denoted by the same reference numerals, and redundant descriptions may be omitted.

[0010] This disclosure describes a motor control device that performs position sensorless vector control of a permanent magnet synchronous motor (PMSM) that drives a compressor as an example. However, the disclosed technology is broadly applicable to motor control devices that estimate the rotor position using magnetic salient polarity for motors having magnetic salient polarity.

[0011] [Example 1] <Motor control device configuration> Figure 1 shows an example configuration of a motor control device according to Embodiment 1 of the present disclosure. In Figure 1, the motor control device 100a includes subtractors 11, 18, 19, a speed controller 12, adders 21, 22, 44, 45, a current command value calculator 14, a current controller 20, a dq / u,v,w converter 23, a PWM (Pulse Width Modulation) modulator 24, and an IPM (Intelligent Power Module) 25. The IPM 25 is connected to a motor M. An example of a motor M is a PMSM.

[0012] Furthermore, the motor control device 100a includes a shunt resistor 26, current sensors 27a and 27b, and a 3φ current calculator 28. However, the motor control device 100a only needs to have either the shunt resistor 26 or the current sensors 27a and 27b.

[0013] Furthermore, the motor control device 100a includes a u,v,w / dq converter 29, an axis error calculator 30, a PLL (Phase Locked Loop) controller 31, a position estimater 32, a 1 / Pn processor 33, and a decoupling controller 36.

[0014] In addition, the motor control device 100a includes a high-frequency voltage command value generator 43, a Pn processor 46, a drive component removal filter 61, a noise reduction command generator 71, a high-frequency current subtractor 41, and a phase generator 42. The high-frequency current subtractor 41 includes subtractors 41a and 41b.

[0015] The subtractor 11 calculates an angular velocity error Δω by subtracting the mechanical angular velocity estimation value ωm, which is the current estimated angular velocity output from the 1 / Pn processor 33, from the mechanical angular velocity command value ωm input from outside the motor control device 100a (for example, a higher-level controller) to the motor control device 100a. Here, the mechanical angular velocity command value ωm * is a command value determined based on the desired rotational speed of the motor M. *

[0016] The speed controller 12 generates a torque command value T such that the angular velocity error Δω approaches 0. *

[0017] The current command value calculator 14 distributes the torque command value T * to the d-axis current command value Id * and the q-axis current command value Iq * on the d-q coordinate axes.

[0018] The subtractor 18 calculates a d-axis current error Id_diff, which is the error between the d-axis current command value Id * and the high-frequency removal d-axis current Idm output from the high-frequency current subtractor 41, by subtracting the high-frequency removal d-axis current Idm from the d-axis current command value Id. The subtractor 19 calculates a q-axis current error Iq_diff, which is the error between the q-axis current command value Iq * and the high-frequency removal q-axis current Iqm output from the high-frequency current subtractor 41, by subtracting the high-frequency removal q-axis current Iqm from the q-axis current command value Iq. * *

[0019] The current controller 20 calculates a provisional d-axis voltage command value Vdt by performing PI (Proportional Integral) control based on the input d-axis current error Id_diff. The current controller 20 also calculates a provisional q-axis voltage command value Vqt by performing PI control based on the input q-axis current error Iq_diff.

[0020] The decoherence controller 36 outputs the electrical angular velocity command value ωe from the Pn processor 46. * The d-axis current command value Id is output from the current command value calculator 14. * Based on this, a d-axis decoupling voltage command value Vda is generated to compensate for the provisional d-axis voltage command value Vdt. In addition, the decoupling controller 36 generates the electrical angular velocity command value ωe output from the Pn processor 46. * The q-axis current command value Iq is output from the current command value calculator 14. * Based on this, a q-axis decoupling voltage command value Vqa is generated to compensate for the provisional q-axis voltage command value Vqt. The d-axis decoupling voltage command value Vda and the q-axis decoupling voltage command value Vqa are decoupling compensation values ​​for feedforward cancellation of interference between the d and q coordinate axes.

[0021] Adder 21 calculates the d-axis drive voltage command value Vdm by adding the d-axis decoupling voltage command value Vda to the provisional d-axis voltage command value Vdt. Adder 22 calculates the q-axis drive voltage command value Vqm by adding the q-axis decoupling voltage command value Vqa to the provisional q-axis voltage command value Vqt. As a result, the d-axis drive voltage command value Vdm and the q-axis drive voltage command value Vqm are obtained, in which interference between the d and q coordinate axes is canceled out by feedforward.

[0022] The dq / u,v,w converter 23 receives the two-phase d-axis voltage command value Vd from the adders 44 and 45. * and q-axis voltage command value Vq * Based on the electrical angular phase θe output from the position estimator 32, the three-phase U-phase output voltage command value Vu * V-phase output voltage command value Vv * and W-phase output voltage command value Vw *Convert to this. The electrical angular phase θe output from the position estimator 32 indicates the current rotor position of the motor M.

[0023] The PWM modulator 24 controls the U-phase output voltage command value Vu * V-phase output voltage command value Vv * W-phase output voltage command value Vw * Based on the PWM carrier signal, a 6-phase PWM signal is generated, and the generated 6-phase PWM signal is output to the IPM25.

[0024] The IPM25 generates three AC voltages (U-phase, V-phase, and W-phase) from a DC voltage Vdc based on the six-phase PWM signal output from the PWM modulator 24, and applies these three AC voltages to the U-phase, V-phase, and W-phase of the motor M.

[0025] The 3φ current calculator 28 calculates the U-phase current Iu, V-phase current Iv, and W-phase current Iw (i.e., the current flowing through the windings of the motor M) of the motor M from the 6-phase PWM switching information output from the PWM modulator 24 and the detected bus current when the bus current is detected using a single-shunt method with a shunt resistor 26. Alternatively, when the U-phase current and V-phase current are detected by the current sensors 27a and 27b, the 3φ current calculator 28 calculates the remaining W-phase current Iw based on Kirchhoff's law "Iu + Iv + Iw = 0". The 3φ current calculator 28 outputs the phase currents Iu, Iv, and Iw of each phase to the u,v,w / dq converter 29.

[0026] The u,v,w / dq converter 29 converts the three-phase U-phase current Iu, V-phase current Iv, and W-phase current Iw into two-phase d-axis current Id and q-axis current Iq based on the electrical angular phase θe output from the position estimator 32.

[0027] The PLL controller 31 calculates the electrical angular velocity ωe, which is the current estimated angular velocity of the motor M, based on the axis error Δθ output from the axis error calculator 30.

[0028] The position estimator 32 estimates the electrical angular phase θe based on the estimated electrical angular velocity ωe.

[0029] The 1 / Pn processor 33 calculates the estimated mechanical angular velocity ωm by dividing the estimated electrical angular velocity ωe by the number of pole pairs Pn of the motor M.

[0030] The Pn processor 46 controls the mechanical angular velocity command value ωm * Multiply the number of pole pairs Pn of motor M by the electrical angular velocity command value ωe * Calculate.

[0031] The high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * This generates the following. In the following, the d-axis high-frequency voltage command value and the q-axis high-frequency voltage command value may be collectively referred to as the "high-frequency voltage vector".

[0032] The phase generator 42 generates a high-frequency voltage with an angular frequency (hereinafter sometimes referred to as "high-frequency angular frequency") ωh * By integrating over the range "0≦ωht≦2π", the high-frequency phase ωht, which is the phase of the high-frequency voltage vector, is generated. High-frequency angular frequency ωh * This is input to the motor control device 100a from an external source (for example, a higher-level controller).

[0033] The high-frequency voltage command value generator 43 receives the application method command signal_noise_method, the high-frequency phase ωht, and the high-frequency voltage amplitude command value Vh from an external source (e.g., a higher-level controller) to the motor control device 100a. * Based on this, the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * Generates a high-frequency voltage vector Vdh. * ,Vqh * This is generated to produce a high-frequency current used for rotor position estimation and does not contribute to torque generation for rotor position estimation. In other words, the high-frequency current used for rotor position estimation is the high-frequency voltage vector Vdh to the motor M. * ,Vqh* It occurs in response to the application of [something].

[0034] The adder 44 controls the d-axis drive voltage command value Vdm and the d-axis high-frequency voltage command value Vdh * By adding these together, the d-axis voltage command value Vd * The adder 45 calculates the q-axis drive voltage command value Vqm and the q-axis high-frequency voltage command value Vqh. * By adding these together, the q-axis voltage command value Vq * The following is calculated. In the following, the d-axis drive voltage command value Vdm and the q-axis drive voltage command value Vqm may be collectively referred to as "drive voltage command value Vm".

[0035] The drive component removal filter 61 controls the mechanical angular velocity command value ωm * and high-frequency angular frequency ωh * Based on this, the high-frequency d-axis current Idh is extracted from the d-axis current Id by removing components other than the high-frequency component from the d-axis current Id. In addition, the drive component removal filter 61 is used with respect to the mechanical angular velocity command value ωm * and high-frequency angular frequency ωh * Based on this, the high-frequency q-axis current Iqh is extracted from the q-axis current Iq by removing components other than the high-frequency component from the q-axis current Iq. Hereafter, components other than the high-frequency component may be referred to as "non-high-frequency current". Non-high-frequency current is the current component generated in conjunction with the driving of the motor M. Hereafter, the high-frequency d-axis current Idh and the high-frequency q-axis current Iqh may be collectively referred to as "high-frequency current Ih".

[0036] The axis error calculator 30 receives the signal_noise_method command, the high-frequency d-axis current Idh, the high-frequency q-axis current Iqh, and the high-frequency angular frequency ωh from the noise reduction command generator 71. * Based on this, the axis error Δθ (the difference between the actual axis of rotation and the estimated axis of rotation) is calculated. In other words, the axis error calculator 30 calculates the high-frequency voltage vector Vdh * ,Vqh * Using the high-frequency current vector generated in response to the application of the current, the axis error Δθ, which is the discrepancy between the dq coordinate axis and the estimated coordinate axis of the dq coordinate axis, is calculated.

[0037] In other words, the position estimator 32, the PLL controller 31, and the axis error calculator 30 estimate the position of the rotor of the motor M based on the high-frequency current generated when a high-frequency voltage is applied to the motor M.

[0038] Subtractor 41a calculates the high-frequency removed d-axis current Idm by subtracting the high-frequency d-axis current Idh from the d-axis current Id. Subtractor 41b calculates the high-frequency removed q-axis current Iqm by subtracting the high-frequency q-axis current Iqh from the q-axis current Iq. In other words, subtractor 41 extracts the drive current of the motor M from the d-axis current Id and q-axis current Iq by subtracting the high-frequency current Ih extracted by the drive component removal filter 61 from the d-axis current Id and q-axis current Iq.

[0039] <Configuration of the noise reduction command generator> Figure 2 shows an example of the configuration of a noise reduction command generator according to Embodiment 1 of this disclosure. The motor control device of this embodiment intermittently applies a high-frequency voltage to the motor M. In Figure 2, the noise reduction command generator 71 has a count calculator 71a and a command generator 71b. The noise reduction command generator 71 sets the timing for applying a high-frequency voltage to estimate the rotor position of the motor M (hereinafter sometimes referred to as "position estimation") and the timing for not applying a high-frequency voltage to reduce noise.

[0040] The count calculator 71a calculates a noise reduction count signal_count based on the high-frequency phase ωht at each predetermined control cycle. Here, the high-frequency phase ωht is repeatedly generated with a cycle of 0 to 2π. The count calculator 71a determines that one cycle of the high-frequency phase ωht has elapsed when the high-frequency phase ωht generated in the current control cycle is less than the high-frequency phase ωht_old generated in the previous control cycle, that is, when the high-frequency phase ωht reaches 2π and returns to 0, and increments the noise reduction count signal_count by 1 (hereinafter sometimes referred to as "incrementing"). Here, when an arbitrary period is set, the noise reduction command generator 71 is set with a number of intervals (hereinafter sometimes referred to as "intervals with application") X in which the high-frequency voltage is applied to the motor M (hereinafter sometimes referred to as "intervals with application") and a number of intervals (hereinafter sometimes referred to as "intervals without application") Y in which the high-frequency voltage is not applied to the motor M (hereinafter sometimes referred to as "intervals without application"), with one interval of the high-frequency phase ωht being considered as one interval. The count calculator 71a increments the noise reduction count signal_count from 1 to X+Y, and resets the noise reduction count signal_count to 1 when it reaches X+Y.

[0041] The command generator 71b generates the signal_noise_method command based on the noise reduction count signal_count. The signal_noise_method command indicates whether or not a high-frequency voltage is applied. When the noise reduction count signal_count is less than or equal to X, the command generator 71b sets the signal_noise_method command to "SIGNAL_ON (0)", while when the noise reduction count signal_count is greater than X, it sets the signal_noise_method command to "SIGNAL_OFF (1)".

[0042] <Operation of the High-Frequency Voltage Command Value Generator> Figure 3 shows an example of the operation of the high-frequency voltage command value generator in Embodiment 1 of this disclosure. Figure 3 shows an example where one period of the high-frequency phase ωht is considered one interval, and the number of applied intervals X is set to "3", while the number of unapplied intervals Y is set to "1". Therefore, the count calculator 71a increments the noise reduction count signal_count from 1 to 4 for each period of the high-frequency phase ωht, and resets the noise reduction count signal_count to 1 when it reaches 4. In addition, the command generator 71b sets the application method command signal_noise_method to "SIGNAL_ON(0)" when the noise reduction count signal_count is 3 or less, while setting the application method command signal_noise_method to "SIGNAL_OFF(1)" when the noise reduction count signal_count is greater than 3.

[0043] The high-frequency voltage command value generator 43 receives the application method command signal_noise_method as input. The high-frequency voltage command value generator 43 generates the high-frequency voltage amplitude command value Vh according to the application method command signal_noise_method, as follows. * Based on the high-frequency phase ωht, the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * Generates.

[0044] When the application method command signal_noise_method input from the noise reduction command generator 71 is set to “SIGNAL_ON(0)”, the high-frequency voltage command value Vdh on the d axis is calculated according to equation (1). * and the q-axis high-frequency voltage command value Vqh * Generates.

number

[0045] On the other hand, when the application method command signal_noise_method input from the noise reduction command generator 71 is set to “SIGNAL_OFF(1)”, the high-frequency voltage command value Vdh of the d axis is calculated according to equation (2) in the high-frequency voltage command value generator 43. * and the q-axis high-frequency voltage command value Vqh * This generates the d-axis high-frequency voltage command value Vdh. In other words, when the application method command signal_noise_method is set to "SIGNAL_OFF(1)", the high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh. * and the q-axis high-frequency voltage command value Vqh * To make it 0, no high-frequency voltage is applied to motor M.

number

[0046] Therefore, when the number of application intervals X is set to "3" and the number of non-application intervals Y is set to "1", as shown in Figure 3, the high-frequency voltage is applied to the motor M in three intervals from the first period T1 to the third period T3, and then not applied to the motor M in one interval of the fourth period T4. Similarly thereafter, the high-frequency voltage is applied to the motor M in three intervals from the fifth period T5 to the seventh period T7, and then not applied to the motor M in one interval of the eighth period T8. Furthermore, the high-frequency voltage is applied to the motor M in three intervals from the ninth period T9 to the eleventh period T11. In this way, the high-frequency voltage command value generator 43 intermittently applies the high-frequency voltage to the motor M. For example, as shown in Figure 3, the high-frequency voltage command value generator 43 provides intervals in which the high-frequency voltage is not applied to the motor M at predetermined time intervals corresponding to three periods of the high-frequency phase ωht. Furthermore, as shown in Figure 3, one period of the high-frequency phase ωht corresponds to one period of the high-frequency voltage. Therefore, the high-frequency voltage command value generator 43 may, for example, provide a section where the high-frequency voltage is not applied to the motor M every three periods of the high-frequency voltage, as shown in Figure 3. Note that the larger the number of applied-voltage sections X is set relative to the number of unapplied-voltage sections Y, the higher the accuracy of position estimation and the less unstable the control of the motor M becomes. On the other hand, the larger the number of unapplied-voltage sections Y is set relative to the number of unapplied-voltage sections X, the greater the effect of suppressing noise. Therefore, the number of applied-voltage sections X and the number of unapplied-voltage sections Y should be set to appropriate values ​​in advance through experiments or other means so as to satisfy the stability of control and the effect of suppressing noise.

[0047] By providing a no-voltage section in this manner, the time-averaged noise level of the motor M can be reduced, thereby suppressing the noise of the motor M.

[0048] <Configuration of the drive component removal filter> Figure 4 shows an example configuration of the drive component removal filter of Embodiment 1 of the present disclosure. In Figure 4, the drive component removal filter 61 includes a high-pass filter 61a, a low-pass filter 61b, a notch filter 61c, and a filter constant calculator 61d.

[0049] Here, the in-phase mirror-phase current vector generator 301 of the axis error calculator 30, which will be described later, functions as a bandpass filter, and in a bandpass filter, the wider the pass frequency bandwidth, the faster the filter's response speed.

[0050] However, due to the sampling rate based on the carrier frequency in the motor control device 100a, it can be difficult to sufficiently separate the frequency band of non-high frequency currents from the frequency band of high frequency currents (Reason 1). For example, since the calculation of the axis error is performed for each period of the carrier signal, the high frequency current frequency fh * If the high frequency (fh) is close to the carrier frequency, the number of samples per period of the high-frequency current decreases, reducing the accuracy of position estimation. Therefore, the high-frequency current frequency fh * It is preferable that the frequency is sufficiently low relative to the carrier frequency (for example, a frequency of 1 / 20th of the carrier frequency).

[0051] Furthermore, since the magnitude of the voltage and current of the drive component that contributes to torque generation changes depending on the driving conditions of the motor M (load, rotational speed, etc.), even under conditions where the voltage and current of the drive component are at their maximum, the drive component frequency band (i.e., the high-frequency current frequency fh) is sufficiently removed to the extent that the current of the drive component (i.e., the non-high-frequency current) is sufficiently removed. * It is necessary to guarantee the attenuation of frequency bands other than the specified band (Reason 2).

[0052] Furthermore, as the motor is driven, a high-frequency current component is generated, mainly due to the motor's cogging torque, corresponding to the motor's rotational speed. Hereafter, the high-frequency current component generated due to the motor's cogging torque will be referred to as the "cogging current." The order of the cogging current frequency is determined by the number of poles on the motor M's rotor (hereinafter sometimes referred to as the "number of rotor poles") and the number of slots on the motor M (hereinafter sometimes referred to as the "number of motor slots"). For example, if motor M is a P-pole, S-slot motor, a cogging current with a frequency equal to the least common multiple of P and S is generated. In addition, cogging currents with frequencies equal to half, twice, three times, ..., n times the least common multiple of P and S are also generated. For example, if motor M is a 6-pole, 9-slot motor, in addition to the 18th-order frequency, which is the least common multiple of 6 and 9, cogging currents with frequencies equal to half the least common multiple of 6 and 9 (the 9th-order frequency), and 18 × n-order frequencies such as the 36th, 54th, ... are generated. Such cogging currents reduce the accuracy of position estimation, so it is necessary to ensure an attenuation level that can sufficiently eliminate the cogging current (Reason 3).

[0053] Furthermore, for the reasons 1, 2, and 3 above, it becomes necessary to narrow the passband of the common-mode mirror-phase current vector generator 301 as a bandpass filter, which slows down the filter response speed of the common-mode mirror-phase current vector generator 301.

[0054] Therefore, in order to input the high-frequency d-axis current Idh and high-frequency q-axis current Iqh to the in-phase mirror-phase current vector generator 301, the drive component removal filter 61 has a high-pass filter 61a, a low-pass filter 61b, a notch filter 61c, and a filter constant calculator 61d, as shown in Figure 4. When the high-frequency d-axis current Idh and high-frequency q-axis current Iqh are input to the in-phase mirror-phase current vector generator 301, the current input to the in-phase mirror-phase current vector generator 301 becomes only the high-frequency current generated by the high-frequency voltage for rotor position estimation. In this way, by combining the high-pass filter 61a, the low-pass filter 61b, and the notch filter 61c to form a band-pass filter, it is possible to widen the passband of the in-phase mirror-phase current vector generator 301 as a band-pass filter, and thus the filter response speed can be increased compared to when a band-pass filter with a narrow passband is used as the in-phase mirror-phase current vector generator 301.

[0055] In Figure 4, the high-pass filter 61a removes the drive current component (hereinafter sometimes referred to as "drive current") Im, which contributes to the torque generation of the motor M, from the d-axis current Id and the q-axis current Iq, thereby obtaining the drive current-removed d-axis current Id1 and drive current-removed q-axis current Iq1. The drive current Im is the current component generated when the motor M is driven by the drive voltage command value Vm. For example, the passband of the high-pass filter 61a is set to the mechanical angular velocity command value ωm * A sufficiently large frequency (for example, the mechanical angular velocity command value ωm) * It is preferable to use a frequency band with a frequency four times or more greater than that of [the specified frequency].

[0056] When the drive current Im is removed by the high-pass filter 61a, a phase shift occurs in the high-frequency currents (drive current removal d-axis current Id1 and drive current removal q-axis current Iq1). Equation (3) shows the first-order phase characteristic θhigh of the high-pass filter 61a. In equation (3), "HIGH_a0" is the filter constant of the high-pass filter 61a. From equation (3), the frequency fh of the high-frequency current generated in the high-pass filter 61a is given by... *The amount of phase shift in the band can be determined.

number

[0057] On the other hand, the high-frequency current frequency fh of the low-pass filter 61b * The first-order phase characteristic θlow of the band is given by equation (4). In equation (4), "LOW_a0" is the filter constant of the low-pass filter 61b.

number

[0058] Therefore, the filter constant calculator 61d calculates the high-frequency angular frequency ωh according to equations (5) and (6). * Based on the filter constant HIGH_a0 of the high-pass filter 61a, the filter constant LOW_a0 of the low-pass filter 61b is calculated. The filter constant LOW_a0 is the high-frequency angular frequency ωh * It is updated as needed in response to changes. The filter constant LOW_a0 calculated by the filter constant calculator 61d is set in the low-pass filter 61b.

number

number

[0059] The low-pass filter 61b performs filtering on the drive current-removed d-axis current Id1 and drive current-removed q-axis current Iq1 using the filter constant LOW_a0 calculated by the filter constant calculator 61d. This compensates for the phase shift that occurs in the drive current-removed d-axis current Id1 and drive current-removed q-axis current Iq1, and removes high-frequency noise Inoise from the drive current-removed d-axis current Id1 and drive current-removed q-axis current Iq1. High-frequency noise Inoise is generated in a frequency band higher than the frequency of the high-frequency current Ih as the motor M is driven. The low-pass filter 61b outputs the filtered noise-removed d-axis current Id2 and noise-removed q-axis current Iq2 to the notch filter 61c.

[0060] The notch filter 61c obtains high-frequency d-axis current Idh and high-frequency q-axis current Iqh by removing the cogging current at cogging frequency fcog from the noise-removed d-axis current Id2 and noise-removed q-axis current Iq2. The cogging frequency fcog is determined by the least common multiple of the rotor pole number P and the motor slot number S and the mechanical angular velocity command value ωm. * The cogging frequency fcog is calculated by dividing the result of multiplication by 2π, and is set as the center frequency of the attenuation region of the notch filter 61c. The cogging frequency fcog is calculated using the mechanical angular velocity command value ωm * This will be updated as changes occur.

[0061] As described above, the drive component removal filter 61 obtains a high-frequency current Ih by removing non-high-frequency currents from the d-axis current Id and the q-axis current Iq. For example, the drive current Im, high-frequency noise Inoise, and cogging current are non-high-frequency currents, which are current components generated in conjunction with the driving of the motor M.

[0062] <Configuration of the axis error calculator> Figure 5 shows an example configuration of the axis error calculator of Embodiment 1 of this disclosure. The axis error calculator 30a shown in Figure 5 corresponds to the axis error calculator 30 shown in Figure 1. In Figure 5, the axis error calculator 30a includes a common-mode mirror-phase current vector generator 301, a mirror-phase estimator 302, a filter-corrected current estimator 303, and a switch 304. The axis error calculator 30a utilizes the magnetic salient polarity of the motor M and calculates the high-frequency angular frequency ωh * Based on the high-frequency d-axis current Idh and the high-frequency q-axis current Iqh, the axis error Δθ is calculated. Figure 6 shows an example configuration of the in-phase mirror-image current vector generator of Embodiment 1 of this disclosure. In Figure 6, the in-phase mirror-image current vector generator 301 has a sign inverter b11 and D-factor filters b12, b13.

[0063] In Figure 6, the sign inverter b11 has a high-frequency angular frequency ωh * The sign is reversed, and the high-frequency angular frequency after the sign reversal is -ωh * Output to the D factor filter b12.

[0064] The D-factor filters b12 and b13 are filters that separate and extract in-phase and enantiomer components, and the D-factor in the D-factor filters b12 and b13 is defined by equation (7) using the identity matrix I, the skew-symmetric matrix J, and the Laplace operator s.

number

[0065] The D-factor filter b12 has a high-frequency angular frequency of -ωh after sign inversion. * Based on this, the common-mode current vector Ihp contained in the high-frequency d-axis current Idh and high-frequency q-axis current Iqh is detected. The common-mode current vector Ihp is the high-frequency voltage vector Vdh * ,Vqh * It rotates in the same direction as the high-frequency magnetic flux vector generated in response to the application of the signal.

[0066] The D-factor filter b13 has a high-frequency angular frequency ωh *Based on this, the mirror phase current vector Ihn contained in the high-frequency d-axis current Idh and high-frequency q-axis current Iqh is detected. The mirror phase current vector Ihn is the high-frequency voltage vector Vdh * ,Vqh * It rotates in the opposite direction to the high-frequency magnetic flux vector generated in response to the application of the signal.

[0067] Here, the D-factor filters b12 and b13 have a frequency response F(s+jωh) for each component of a 2×1 vector, which is a scalar signal. * It functions equivalently to the filter of ). Therefore, by designing F(s) to have low-pass characteristics and applying it to D-factor filters b12 and b13, the D-factor filters b12 and b13 become ωh * This will function as a bandpass filter with a center frequency of . Furthermore, the D-factor filters b12 and b13 have polarity separation characteristics. In other words, the polarity separation bandpass characteristics of the D-factor filters b12 and b13 allow the high-frequency d-axis current Idh and high-frequency q-axis current Iqh to be separated into the common-mode current vector Ihp and the mirror-mode current vector Ihn.

[0068] As described above, the in-phase mirror phase current vector generator 301 calculates the in-phase current vector Ihp and the mirror phase current vector Ihn using high-frequency current, whereas the filter-corrected current estimator 303 calculates the in-phase current vector Ihp and the mirror phase current vector Ihn without using high-frequency current, as follows. The filter-corrected current estimator 303 estimates the in-phase current vector Ihp according to equation (8) and the mirror phase current vector Ihn according to equation (9). In equations (8) and (9), Ld is the d-axis inductance of the motor M, Lq is the q-axis inductance of the motor M, R(2Δθ) is a rotation matrix with a rotation angle of 2Δθ, up(ωht) is a 2×1 unit positive-sequence matrix represented by equation (10), and un(ωht) is a 2×1 unit negative-sequence matrix represented by equation (11). Also, Δθ in equation (9) is a provisional axis error calculated by an extended induced voltage method or the like, which uses the induced voltage generated by the rotation of the motor M to estimate the position. In other words, the filter-corrected current estimator 303 estimates the common-mode current vector Ihp and the mirror-mode current vector Ihn that occur when a high-frequency voltage is applied to the motor, by using theoretical formulas.

number

number

number

number

[0069] Here, in the no-applied section, the high-frequency d-axis current Idh and high-frequency q-axis current Iqh do not flow, making it difficult to calculate the common-mode current vector Ihp and mirror-phase current vector Ihn using the common-mode mirror-phase current vector generator 301. Therefore, the switch 304 switches the output source of the common-mode current vector Ihp and mirror-phase current vector Ihn input to the mirror-phase estimator 302 between the common-mode mirror-phase current vector generator 301 and the filter-corrected current estimator 303, according to the application method command signal_noise_method input from the noise reduction command generator 71. When the application method command signal_noise_method is set to "SIGNAL_ON(0)", the switch 304 connects the common-mode mirror-phase current vector generator 301 to the mirror-phase estimator 302, thereby making the output source of the common-mode current vector Ihp and mirror-phase current vector Ihn input to the mirror-phase estimator 302 the common-mode mirror-phase current vector generator 301. On the other hand, when the application method command signal_noise_method is set to “SIGNAL_OFF(1)”, the switch 304 connects the filter-corrected current estimator 303 to the mirror phase estimator 302, thereby making the filter-corrected current estimator 303 the source of the common-mode current vector Ihp and mirror phase current vector Ihn input to the mirror phase estimator 302. In this way, in the applied-current section, the common-mode current vector Ihp and mirror phase current vector Ihn generated by the common-mode mirror phase current vector generator 301 are input to the mirror phase estimator 302, while in the unapplied-current section, the common-mode current vector Ihp and mirror phase current vector Ihn estimated by the filter-corrected current estimator 303 are input to the mirror phase estimator 302. Therefore, even when an unapplied-current section is provided to suppress the noise of the motor M, the common-mode current vector Ihp and mirror phase current vector Ihn can be calculated, preventing the control of the motor M from becoming unstable.

[0070] The mirror phase estimator 302 calculates the axis error Δθ according to equations (12) and (13) using the common-mode current vector Ihp and the mirror phase current vector Ihn input from the common-mode mirror phase current vector generator 301 or the filter-corrected current estimator 303. In other words, the mirror phase estimator 302 calculates the axis error Δθ as the arctangent of the composite vector obtained by vector addition of the common-mode current vector Ihp and the mirror phase current vector Ihn, which have the same norm.

number

number

[0071] Furthermore, the switch 304 may connect the in-phase mirror phase current vector generator 301 to the mirror phase estimator 302 in the section from the time when the application method command signal_noise_method set to “SIGNAL_ON(0)” is input started (hereinafter sometimes referred to as the “start time of the applied-voltage section”) after a predetermined waiting time has elapsed, until the time when the application method command signal_noise_method set to “SIGNAL_OFF(1)” is input started (hereinafter sometimes referred to as the “end time of the applied-voltage section”) (hereinafter sometimes referred to as the “first predetermined section”), and connect the filter correction current estimator 303 to the mirror phase estimator 302 in the section from the end time of the applied-voltage section until a predetermined waiting time has elapsed from the start time of the next applied-voltage section (hereinafter sometimes referred to as the “second predetermined section”). Here, immediately after the application of the high-frequency voltage, the high-frequency current becomes unstable. Therefore, the predetermined waiting time is set to the time until the high-frequency current stabilizes, for example, a time corresponding to one cycle of the high-frequency voltage. In this case, as explained in Figure 3, from the first period T1 to the fifth period T5, the start of the first period T1 and the fifth period T5 correspond to the start of the applied voltage section. Also, the first period T1 and the fifth period T5 correspond to a predetermined waiting time. Furthermore, the end of the first period T1 and the fifth period T5 corresponds to the time when the predetermined waiting time has elapsed. Also, the end of the third period T3 corresponds to the end of the applied voltage section. Furthermore, the second period T2 to the third period T3 becomes the first predetermined section, and the first period T1 and the fourth period T4 to the fifth period T5 become the second predetermined section. By doing this, it is possible to prevent the control of the motor M from becoming unstable in the section immediately after the application of the high-frequency voltage when the high-frequency current is not yet stable.

[0072] The above describes Example 1.

[0073] [Example 2] <Operation of the count calculator, command generator, high-frequency voltage command value generator, and switch> Figure 7 shows an example of the operation of the count calculator, command generator, and high-frequency voltage command value generator in Embodiment 2 of this disclosure. Embodiment 2 differs from Embodiment 1 in that, as shown in Figure 7, there is no no-voltage interval, and the phase of the high-frequency voltage is alternately inverted between a first phase, which is in positive phase, and a second phase, which is inverse phase, which is the opposite phase to the first phase. For example, the amplitude of the high-frequency voltage in the first phase and the amplitude of the high-frequency voltage in the second phase are the same. The differences from Embodiment 1 will be explained below.

[0074] The count calculator 71a calculates the noise reduction count signal_count based on the high-frequency phase ωht. The count calculator 71a determines that one period of the high-frequency phase ωht has elapsed if the high-frequency phase ωht generated in the current control cycle is less than the high-frequency phase ωht_old generated in the previous control cycle, and increments the noise reduction count signal_count. Here, when an arbitrary period is set, the noise reduction command generator 71 is set to invert the phase of the high-frequency voltage every X periods, with one period of the high-frequency phase ωht being considered as one period. The count calculator 71a increments the noise reduction count signal_count from 1 to X, and resets the noise reduction count signal_count to 1 when it reaches X.

[0075] The command generator 71b generates an application method command signal_noise_method based on the noise reduction count signal_count. The application method command signal_noise_method indicates whether the phase of the high-frequency voltage is first phase or second phase. Each time the noise reduction count signal_count reaches X, the command generator 71b switches the setting of the application method command signal_noise_method between "SIGNAL_ON_POS(0)" and "SIGNAL_ON_NEG(1)". That is, if the application method command signal_noise_method is set to "SIGNAL_ON_POS(0)" when the noise reduction count signal_count reaches X, the command generator 71b changes the application method command signal_noise_method from "SIGNAL_ON_POS(0)" to "SIGNAL_ON_NEG(1)". On the other hand, if the noise reduction count signal_count reaches X and the application method command signal_noise_method is set to "SIGNAL_ON_NEG(1)", the command generator 71b changes the application method command signal_noise_method from "SIGNAL_ON_NEG(1)" to "SIGNAL_ON_POS(0)".

[0076] Figure 7 shows an example where one period of the high-frequency phase ωht is defined as one interval, the number of intervals X is set to "3", and the phase of the high-frequency voltage is inverted every three intervals. Therefore, the count calculator 71a increments the noise reduction count signal_count from 1 to 3 for each period of the high-frequency phase ωht, and resets the noise reduction count signal_count to 1 when it reaches 3. In addition, the command generator 71b changes the application method command signal_noise_method, which is set to "SIGNAL_ON_POS(0)", to "SIGNAL_ON_NEG(1)", and changes the application method command signal_noise_method, which is set to "SIGNAL_ON_NEG(1)", back to "SIGNAL_ON_POS(0)" each time the noise reduction count signal_count reaches 3.

[0077] The high-frequency voltage command value generator 43 receives the application method command signal_noise_method as input. The high-frequency voltage command value generator 43 generates the high-frequency voltage amplitude command value Vh according to the application method command signal_noise_method, as follows. * Based on the high-frequency phase ωht, the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * Generates.

[0078] When the application method command signal_noise_method input from the noise reduction command generator 71 is set to “SIGNAL_ON_POS(0)”, the high-frequency voltage command value Vdh on the d axis is calculated according to equation (1). * and the q-axis high-frequency voltage command value Vqh * Generates.

[0079] On the other hand, when the application method command signal_noise_method input from the noise reduction command generator 71 is set to “SIGNAL_ON_NEG(1)”, the high-frequency voltage command value Vdh of the d axis is calculated according to equation (14) in the high-frequency voltage command value generator 43. *and the q-axis high-frequency voltage command value Vqh * This generates a high-frequency voltage. In other words, when the signal_noise_method command is set to "SIGNAL_ON_NEG(1)", the high-frequency voltage command value generator 43 applies a high-frequency voltage to the motor M that is in the opposite phase to the high-frequency voltage applied to the motor M when the signal_noise_method command is set to "SIGNAL_ON_NEG(1)".

number

[0080] Therefore, when the number of intervals X is set to "3", as shown in Figure 7, a positive-sequence high-frequency voltage is applied to the motor M in three intervals from the first period T1 to the third period T3, and then an inverse-sequence high-frequency voltage is applied to the motor M in three intervals from the fourth period T4 to the sixth period T6. After the inverse-sequence high-frequency voltage is applied to the motor M in three intervals from the fourth period T4 to the sixth period T6, a positive-sequence high-frequency voltage is applied to the motor M in three intervals from the seventh period T7 to the ninth period T9. In this way, the high-frequency voltage command value generator 43 alternately applies a first-phase high-frequency voltage and a high-frequency voltage that is in the opposite phase to the first-phase high-frequency voltage to the motor M as high-frequency voltages.

[0081] In this way, by alternately providing sections in which a first-phase high-frequency voltage is applied to the motor M and sections in which a second-phase voltage, which has the opposite phase to the first phase, is applied to the motor M, the time-averaged noise level of the motor M can be reduced, thereby suppressing the noise of the motor M. Furthermore, since there are no sections without voltage application, a decrease in the accuracy of position estimation can be prevented.

[0082] Furthermore, even when the application method command signal_noise_method is set to either "SIGNAL_ON_POS(0)" or "SIGNAL_ON_NEG(1)", the switch 304 connects the common-mode mirror phase current vector generator 301 to the mirror phase estimator 302, thereby making the common-mode mirror phase current vector Ihp and mirror phase current vector Ihn input to the mirror phase estimator 302 the source of output for the common-mode mirror phase current vector generator 301.

[0083] Furthermore, the switch 304 may connect the in-phase mirror phase current vector generator 301 to the mirror phase estimator 302 in the section from the time when the application method command signal_noise_method switches between "SIGNAL_ON_POS(0)" and "SIGNAL_ON_NEG(1)" (hereinafter sometimes referred to as the "first phase switching time") until a predetermined waiting time has elapsed, to the next phase switching time (hereinafter sometimes referred to as the "second phase switching time") (hereinafter sometimes referred to as the "third predetermined section"), and connect the filter correction current estimator 303 to the mirror phase estimator 302 in the section from the second phase switching time until a predetermined waiting time has elapsed (hereinafter sometimes referred to as the "fourth predetermined section"). Here, immediately after the phase switching of the high-frequency voltage, the high-frequency current becomes unstable. Therefore, the predetermined waiting time is set to the time until the high-frequency current stabilizes, for example, a time corresponding to one cycle of the high-frequency voltage. In this case, as explained in Figure 7 for the first period T1 to the fifth period T5, the start of the first period T1 and the fourth period T4 correspond to the first phase switching point. The first period T1 and the fourth period T4 correspond to a predetermined waiting time. The end of the first period T1 and the fourth period T4 corresponds to the point when the predetermined waiting time has elapsed. The end of the third period T3 corresponds to the second phase switching point. The second period T2 to the third period T3 constitute the third predetermined section, and the first period T1 and the fourth period T4 to the fifth period T5 constitute the fourth predetermined section. This prevents the control of the motor M from becoming unstable in the section immediately after the phase switching of the high-frequency voltage when the high-frequency current is not yet stable.

[0084] The above describes Example 2.

[0085] [Example 3] <Operation of the count calculator, command generator, high-frequency voltage command value generator, and switch> Figure 8 shows an example of the operation of the count calculator, command generator, and high-frequency voltage command value generator in Embodiment 3 of this disclosure. Embodiment 3 differs from Embodiments 1 and 2 in that, as shown in Figure 8, it alternately inverts the phase of the high-frequency voltage between a first phase, which is in the positive phase, and a second phase, which is in the opposite phase to the first phase, while providing a no-voltage section. In other words, Embodiment 3 is a combination of Embodiments 1 and 2. The differences from Embodiments 1 and 2 will be explained below.

[0086] The count calculator 71a calculates the noise reduction count signal_count based on the high-frequency phase ωht. The count calculator 71a determines that one period of the high-frequency phase ωht has elapsed if the high-frequency phase ωht generated in the current control cycle is less than the high-frequency phase ωht_old generated in the previous control cycle, and increments the noise reduction count signal_count. Here, when an arbitrary period is set, one period of the high-frequency phase ωht is considered as one interval, and the number of intervals without application Y is set in the noise reduction command generator 71 for the number of intervals with application X. The count calculator 71a increments the noise reduction count signal_count from 1 to X+Y, and resets the noise reduction count signal_count to 1 when it reaches X+Y.

[0087] The command generator 71b generates the signal_noise_method application method command based on the noise reduction count signal_count. The signal_noise_method application method command indicates whether or not a high-frequency voltage is applied and whether the phase of the high-frequency voltage is first phase or second phase. When the noise reduction count signal_count is greater than X, the command generator 71b sets the signal_noise_method application method command signal_noise_method to "SIGNAL_OFF(2)". On the other hand, when the noise reduction count signal_count is less than or equal to X, the command generator 71b sets the signal_noise_method application method command signal_noise_method to "SIGNAL_ON_POS(0)" or "SIGNAL_ON_NEG(1)". Furthermore, the command generator 71b switches the setting of the application method command signal_noise_method between "SIGNAL_ON_POS(0)" and "SIGNAL_ON_NEG(1)" each time the noise reduction count signal_count reaches X+Y.

[0088] Figure 8 shows an example where one period of the high-frequency phase ωht is considered one interval, and the number of applied intervals X is set to "3", while the number of unapplied intervals Y is set to "1". Therefore, the count calculator 71a increments the noise reduction count signal_count from 1 to 4 for each period of the high-frequency phase ωht, and resets the noise reduction count signal_count to 1 when it reaches 4. In addition, the command generator 71b sets the application method command signal_noise_method to "SIGNAL_ON_POS(0)" or "SIGNAL_ON_NEG(1)" when the noise reduction count signal_count is 3 or less, while setting the application method command signal_noise_method to "SIGNAL_OFF(2)" when the noise reduction count signal_count is greater than 3. Furthermore, when the noise reduction count signal_count is 3 or less, the command generator 71b sets the application method command signal_noise_method to "SIGNAL_ON_POS(0)" or "SIGNAL_ON_NEG(1)". However, the application method command signal_noise_method set at this time is alternately set between "SIGNAL_ON_POS(0)" and "SIGNAL_ON_NEG(1)" each time the noise reduction count signal_count is reset to 1.

[0089] The high-frequency voltage command value generator 43 receives the application method command signal_noise_method as input. The high-frequency voltage command value generator 43 generates the high-frequency voltage amplitude command value Vh according to the application method command signal_noise_method, as follows. * Based on the high-frequency phase ωht, the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * Generates.

[0090] When the applied method command signal_noise_method input from the noise reduction command generator 71 is set to "SIGNAL_ON_POS(0)", the high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * in accordance with Equation (1).

[0091] Also, when the applied method command signal_noise_method input from the noise reduction command generator 71 is set to "SIGNAL_ON_NEG(1)", the high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * in accordance with Equation (14).

[0092] Also, when the applied method command signal_noise_method input from the noise reduction command generator 71 is set to "SIGNAL_OFF(2)", the high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * in accordance with Equation (2).

[0093] Therefore, when the number of applied intervals X is set to "3" while the number of non-applied intervals Y is set to "1", as shown in FIG. 8, after a positive-phase high-frequency voltage is applied to the motor M in three intervals of the first cycle T1 to the third cycle T3, the high-frequency voltage is not applied to the motor M in one interval of the fourth cycle T4. Also, after a negative-phase high-frequency voltage is applied to the motor M in three intervals of the fifth cycle T5 to the seventh cycle T7, the high-frequency voltage is not applied to the motor M in one interval of the eighth cycle T8. Also, a positive-phase high-frequency voltage is applied to the motor M in three intervals of the ninth cycle T9 to the eleventh cycle T11. Thus, while providing non-applied intervals, the high-frequency voltage command value generator 43 alternately applies, as the high-frequency voltage, a high-frequency voltage of the first phase and a high-frequency voltage having a phase opposite to that of the high-frequency voltage of the first phase to the motor M.

[0094] In this way, by alternately providing sections where a first-phase high-frequency voltage is applied to the motor M, and sections where a second-phase voltage, which is in the opposite phase to the first phase, is applied to the motor M, while also providing a section where no voltage is applied, the time-averaged noise level of the motor M can be reduced, thereby suppressing the noise of the motor M.

[0095] When the application method command signal_noise_method is set to "SIGNAL_ON_POS(0)" or "SIGNAL_ON_NEG(1)", the switch 304 connects the common-mode mirror phase current vector generator 301 to the mirror phase estimator 302, thereby making the common-mode current vector Ihp and mirror phase current vector Ihn input to the mirror phase estimator 302 the source of the output. On the other hand, when the application method command signal_noise_method is set to "SIGNAL_OFF(2)", the switch 304 connects the filter-corrected current estimator 303 to the mirror phase estimator 302, thereby making the common-mode current vector Ihp and mirror phase current vector Ihn input to the mirror phase estimator 302 the source of the output.

[0096] In addition, the switch 304 may, as in Embodiment 1, connect the in-phase mirror phase current vector generator 301 to the mirror phase estimator 302 in the section from when a predetermined waiting time has elapsed from the time when the application method command signal_noise_method set to “SIGNAL_ON_POS(0)” or “SIGNAL_ON_NEG(1)” is input (i.e., the start of the applied voltage section) until when the application method command signal_noise_method set to “SIGNAL_OFF(2)” is input (i.e., the end of the applied voltage section) (hereinafter sometimes referred to as the “fifth predetermined section”), and connect the filter correction current estimator 303 to the mirror phase estimator 302 in the section from the end of the applied voltage section until a predetermined waiting time has elapsed from the start of the next applied voltage section (hereinafter sometimes referred to as the “sixth predetermined section”). Here, immediately after the application of the high-frequency voltage, the high-frequency current becomes unstable. Therefore, the predetermined waiting time is set to the time until the high-frequency current stabilizes, for example, a time corresponding to one cycle of the high-frequency voltage. In this case, as explained in Figure 8, the start of the first period T1 to the fifth period T5 corresponds to the start of the applied voltage section. The first period T1 and the fifth period T5 correspond to the predetermined waiting time. The end of the first period T1 and the fifth period T5 corresponds to the time when the predetermined waiting time has elapsed. The end of the third period T3 corresponds to the end of the applied voltage section. The second period T2 to the third period T3 becomes the fifth predetermined section, and the first period T1 and the fourth period T4 to the fifth period T5 become the sixth predetermined section. This prevents the control of the motor M from becoming unstable in the section immediately after the application of the high-frequency voltage when the high-frequency current is not yet stable.

[0097] Further, in the non-application period, the filter correction current estimator 303 may calculate the in-phase current vector Ihp and the mirror-image current vector Ihn based on the next-applied high-frequency voltage between the high-frequency voltage of the first phase and the high-frequency voltage of the second phase as the high-frequency voltage applied to the motor M. By doing so, at the start of the next application period, the changes in the in-phase current vector Ihp and the mirror-image current vector Ihn input to the mirror-image estimator 302 are reduced, so that it is possible to prevent the control of the motor M from becoming unstable.

[0098] The above is the description of Example 3.

[0099] [Example 4] <Operation of High-Frequency Voltage Command Value Generator> FIG. 9 is a diagram showing an operation example of the high-frequency voltage command value generator according to Example 4 of the present disclosure. In Example 4, as shown in FIG. 9, at the end of the application period, the q-axis high-frequency voltage command value Vqh * is generated longer by a quarter cycle than the d-axis high-frequency voltage command value Vdh * , which is different from Example 3. Hereinafter, the differences from Example 3 will be described.

[0100] When the application method command signal_noise_method input from the noise reduction command generator 71 to the high-frequency voltage command value generator 43 is set to "SIGNAL_OFF(2)", the high-frequency voltage command value generator 43 generates the d-axis high-frequency voltage command value Vdh * and the q-axis high-frequency voltage command value Vqh * according to Equation (2). However, the high-frequency voltage command value generator 43 continues to generate the q-axis high-frequency voltage command value Vqh * according to Equation (1) or Equation (14) for a quarter cycle from the end of the application period. Specifically, when the application method command signal_noise_method was set to "SIGNAL_ON_POS(0)" in the immediately preceding application period, the high-frequency voltage command value generator 43 generates the q-axis high-frequency voltage command value Vqh *The generation continues, and if the application method command signal_noise_method was set to “SIGNAL_ON_NEG(1)” in the immediately preceding applied section, the q-axis high-frequency voltage command value Vqh is determined according to equation (14). * Continue generating.

[0101] Therefore, when the number of applied intervals X is set to "3" and the number of unapplied intervals Y is set to "1", as shown in Figure 9, a positive-sequence high-frequency voltage is applied to the motor M in three intervals from the first period T1 to the third period T3, and then no high-frequency voltage is applied to the motor M in one interval, the fourth period T4. Also, an inverse-sequence high-frequency voltage is applied to the motor M in three intervals from the fifth period T5 to the seventh period T7, and then no high-frequency voltage is applied to the motor M in one interval, the eighth period T8. Furthermore, a positive-sequence high-frequency voltage is applied to the motor M in three intervals from the ninth period T9 to the eleventh period T11. However, in the unapplied intervals of the fourth period T4 and the eighth period T8, for a period of one-quarter of a cycle from the start of the unapplied interval (i.e., the end of the applied interval), only the q-axis high-frequency voltage is applied to the motor M, out of the d-axis high-frequency voltage and q-axis high-frequency voltage.

[0102] By doing so, as shown in Figure 9, both the high-frequency d-axis current Idh and the high-frequency q-axis current Iqh can be kept near zero in the no-voltage section. This suppresses abrupt changes in the current input to the current controller 20, preventing the control of the motor M from becoming unstable.

[0103] Furthermore, at the end of the applied voltage section, the d-axis high-frequency voltage command value Vdh * The q-axis high-frequency voltage command value Vqh is better than the q-axis high-frequency voltage command value Vqh * The point of generating it with a period one-quarter longer may also be applied to Example 1.

[0104] The above describes Example 4.

[0105] [Example 5] <Configuration of the axis error calculator> Figure 10 shows an example of the configuration of the axis error calculator in Embodiment 5 of this disclosure. The axis error calculator 30b shown in Figure 10 corresponds to the axis error calculator 30 shown in Figure 1. In Figure 10, the axis error calculator 30b includes a common-mode mirror phase current vector generator 301, a mirror phase estimator 302, a filter correction current estimator 305, and a switch 304. The differences from Embodiment 1 will be described below.

[0106] The filter-corrected current estimator 305 estimates the common-mode current vector Ihp according to equation (15) and the mirror phase current vector Ihn according to equation (16). Here, the filter-corrected current estimator 305 calculates the common-mode current vector Ihp and the mirror phase current vector Ihn for each period of the high-frequency voltage, and in equations (15) and (16), Ihp peak Ihn is the peak value of the in-mode current vector Ihp in the previous period. peak θi is the peak value of the mirror phase current vector Ihn in the previous period. dp is the current phase. Also, up(ωht+θi dp ) is a 2×1 identity positive-sequence matrix represented by equation (17), un(ωht+θi dp ) is a 2 × 1 unit inverse phase matrix represented by equation (18). Also, Δθ in equation (16) is a provisional axis error calculated by an extended induced voltage method or the like, which uses the induced voltage generated by the rotation of the motor M to estimate the position. In other words, the filter-corrected current estimator 303 estimates the common-mode current vector Ihp and mirror-mode current vector Ihn that are generated when a high-frequency voltage is applied to the motor in the current period by using the peak values ​​of the common-mode current vector Ihp and mirror-mode current vector Ihn in the previous period.

number

number

number

number

[0107] The above describes Example 5.

[0108] [Example 6] <Motor control device configuration> Figure 11 shows an example configuration of a motor control device according to Embodiment 6 of this disclosure. The motor control device 100b shown in Figure 11 differs from the device 100a of Embodiment 1 in that it has a drive / harmonic separation filter 62 instead of a high-frequency current subtractor 41 and a drive component removal filter 61. The differences from Embodiment 1 will be explained below.

[0109] The drive / harmonic separation filter 62 receives the d-axis current Id and the q-axis current Iq from the u,v,w / dq converter 29. The drive / harmonic separation filter 62, for example, similar to the drive component removal filter 61 in Example 1, uses a high-pass filter, a low-pass filter, and a notch filter to remove non-high-frequency currents from the d-axis current Id to obtain a high-frequency d-axis current Idh, and also removes non-high-frequency currents from the q-axis current Iq to obtain a high-frequency q-axis current Iqh. Furthermore, the drive / harmonic separation filter 62, for example, uses a notch filter to remove the high-frequency d-axis current Idh from the d-axis current Id to obtain a high-frequency removed d-axis current Idm, and also removes the high-frequency removed q-axis current Iqh from the q-axis current Iq to obtain a high-frequency removed q-axis current Iqm. The drive / harmonic separation filter 62 outputs the separated high-frequency d-axis current Idh and high-frequency q-axis current Iqh to the axis error calculator 30, and outputs the separated high-frequency removed d-axis current Idm and high-frequency removed q-axis current Iqm to the subtractors 18 and 19.

[0110] The above describes Example 6.

[0111] [Example 7] <Motor control device configuration> Figure 12 shows an example of the configuration of a motor control device according to Embodiment 7 of this disclosure. The motor control device 100c shown in Figure 12 differs from the device 100a of Embodiment 1 in that it has a drive current subtractor 48 and a high-frequency current removal filter 63 instead of a high-frequency current subtractor 41 and a drive component removal filter 61. The differences from Embodiment 1 will be explained below.

[0112] The drive current subtractor 48 has subtractors 48a and 48b. The harmonic current rejection filter 63 receives the d-axis current Id and the q-axis current Iq from the u,v,w / dq converter 29. Additionally, the d-axis current Id is input to subtractor 48a from the u,v,w / dq converter 29, and the q-axis current Iq is input to subtractor 48b from the u,v,w / dq converter 29.

[0113] The high-frequency current removal filter 63 obtains a high-frequency removed d-axis current Idm by removing the high-frequency d-axis current Idh from the d-axis current Id using, for example, a notch filter, and also obtains a high-frequency removed q-axis current Iqm by removing the high-frequency q-axis current Iqh from the q-axis current Iq. The high-frequency current removal filter 63 outputs the high-frequency removed d-axis current Idm to subtractors 18 and 48a, and the high-frequency removed q-axis current Iqm to subtractors 19 and 48b.

[0114] Subtractor 48a calculates the high-frequency d-axis current Idh by subtracting the high-frequency removed d-axis current Idm from the d-axis current Id. Subtractor 48b calculates the high-frequency q-axis current Iqh by subtracting the high-frequency removed q-axis current Iqm from the q-axis current Iq.

[0115] The above describes Example 7.

[0116] Examples 6 and 7 describe a method different from Example 1 for separating the d-axis current Id into a high-frequency d-axis current Idh and a high-frequency removed d-axis current Idm, and for separating the q-axis current Iq into a high-frequency q-axis current Iqh and a high-frequency removed q-axis current Iqm. Here, the high-frequency voltage is applied intermittently, and the application and non-application periods are repeatedly switched. Furthermore, the digital filters such as notch filters used to remove the high-frequency current in Examples 6 and 7 perform filtering using past input and output values. Therefore, if the high-frequency currents included in the d-axis current Id and q-axis current Iq, which are the input values ​​of the filter, change rapidly when the application and non-application periods are switched, the filter output becomes unstable. In addition, in Example 6, since filters are used for both the extraction of the high-frequency d-axis current Idh and high-frequency q-axis current Iqh, and the extraction of the high-frequency removed d-axis current Idm and high-frequency removed q-axis current Iqm, the processing time becomes longer. On the other hand, the method of Example 1 removes the continuously applied non-high frequency current by the drive component removal filter 61 to extract the high frequency d-axis current Idh and high frequency q-axis current Iqh. Then, the high frequency removed d-axis current Idm is calculated by subtracting the high frequency d-axis current Idh from the d-axis current Id using the high frequency current subtractor 41, and the high frequency removed q-axis current Iqm is calculated by subtracting the high frequency q-axis current Iqh from the q-axis current Iq. Therefore, the control of the motor M is most stable according to the method of Example 1.

[0117] [Example 8] <Motor control device configuration> Figure 13 shows an example of the configuration of a motor control device according to Embodiment 8 of this disclosure. Compared to the motor control device 100a of Embodiment 1 (Figure 1), the motor control device 100d shown in Figure 13 has a high-frequency calculator 81 and an application method determination device 82. In the motor control device 100d, the noise reduction command generator 71 switches the application method of the high-frequency voltage based on the determination result of the application method determination device 82, which is different from Embodiment 1. The differences from Embodiment 1 will be explained below.

[0118] <Processing procedure in a high-frequency calculator> Figure 14 is a flowchart showing an example of the processing procedure in the high-frequency calculator of Embodiment 8 of this disclosure.

[0119] The order of the fluctuation frequency of current fluctuations caused by cogging torque is determined by the number of rotor poles and the number of motor slots. For example, if motor M is a P-pole, S-slot motor, current fluctuations occur at a frequency that is the least common multiple of P and S. Furthermore, current fluctuations also occur at frequencies that are half, twice, three times, ..., n times the least common multiple of P and S. For example, if motor M is a 6-pole, 9-slot motor, current fluctuations occur at the 18th order frequency, which is the least common multiple of 6 and 9, as well as at the 9th order frequency, which is half the least common multiple of 6 and 9, and at 18 × n order frequencies such as the 36th, 54th, ...

[0120] On the other hand, the common-mode mirror-phase current vector generator 301 functions as a bandpass filter, and the wider the pass frequency range, the faster the response speed of the common-mode mirror-phase current vector generator 301 as a bandpass filter. For this reason, when designing the filter constants in the common-mode mirror-phase current vector generator 301, it is preferable to design the filter constants to be sufficient to remove the driving component current.

[0121] Here, the axis error calculation is performed for each cycle of the carrier signal, therefore the high-frequency current frequency fh * If the high frequency (fh) is close to the carrier frequency, the number of samples per period of the high-frequency current decreases, reducing the accuracy of position estimation. Therefore, the high-frequency current frequency fh * It is preferable that the frequency is sufficiently low (e.g., 1 / 20th) relative to the carrier frequency. Also, since the driving component current is removed by the in-mode mirror phase current vector generator 301 which functions as a bandpass filter, the high-frequency current frequency fh * It is preferable that the frequency is sufficiently higher (e.g., 25 times) than the frequency band of the drive voltage command value. Therefore, the high-frequency current frequency fh * It is preferable that the frequency is set to a frequency that is sufficiently lower than the carrier frequency of the motor control device 100d and sufficiently higher than the frequency band of the drive voltage command value.

[0122] However, high-frequency current frequency fh * If we keep it a fixed value, the high-frequency angular frequency ωh * This also becomes a fixed value, and the pass-through frequency range of the common-mode mirror-phase current vector generator 301 is fixed. Therefore, depending on the rotational speed of the motor M, fluctuation frequencies caused by cogging torque may be included in the pass-through frequency range of the common-mode mirror-phase current vector generator 301. In this case, the high-frequency voltage vector Vdh in the common-mode mirror-phase current vector generator 301 does not contribute to torque generation for rotor position estimation. * ,Vqh * Components other than the current generated by the motor are also extracted. As a result, the accuracy of position estimation deteriorates depending on the rotation speed of motor M.

[0123] Therefore, in step S100, the high-frequency calculator 81 calculates the least common multiple of the rotor pole number P and the motor slot number S and the mechanical angular velocity command value ωm according to equation (19). * By dividing the result of multiplying by 2π, the cogging frequency fcog, which represents the fluctuation frequency, is calculated.

number

[0124] Next, in step S105, the high-frequency calculator 81 determines, according to equation (20), whether the sum of the cogging frequency fcog and the first margin fmg1 (hereinafter sometimes referred to as the "margin sum frequency") is greater than or equal to a predetermined reference frequency fh1. The reference frequency fh1 is the high-frequency current frequency fh *A preferred frequency is predetermined, that is, a frequency that is sufficiently lower than the carrier frequency of the motor control device 100d and sufficiently higher than the frequency band of the drive voltage command value. For example, when the carrier frequency is 4kHz and the drive voltage frequency is 8Hz, the reference frequency fh1 can be set to 200Hz. If the margin addition frequency is greater than or equal to the reference frequency fh1 (step S105: Yes), the process proceeds to step S110, and if the margin addition frequency is less than the reference frequency fh1 (step S105: No), the process proceeds to step S115.

number

[0125] In step S110, the high-frequency calculator 81 calculates the margin summing frequency according to equation (21) to the high-frequency current frequency fh * Set as and the high-frequency current frequency fh after setting * The output is sent to the application method determination device 82.

number

[0126] On the other hand, in step S115, the high-frequency calculator 81 determines whether the value obtained by subtracting the second margin fmg2 from twice the cogging frequency fcog (hereinafter sometimes referred to as the "margin subtraction frequency") is less than the reference frequency fh1, according to equation (22). If the margin subtraction frequency is less than the reference frequency fh1 (step S115: Yes), the process proceeds to step S120; if the margin subtraction frequency is greater than or equal to the reference frequency fh1 (step S115: No), the process proceeds to step S125.

number

[0127] The first margin fmg1 is a margin predetermined based on the width of the pass frequency region of the common-mode mirror-phase current vector generator 301, the magnitude of the high-frequency current, and the magnitude of the fluctuation of the cogging frequency fcog, for example, 60 Hz. The second margin fmg2 is a margin predetermined based on the width of the pass frequency region of the common-mode mirror-phase current vector generator 301, the magnitude of the high-frequency current, and the magnitude of the fluctuation at twice the frequency of the cogging frequency fcog, for example, 30 Hz.

[0128] In step S120, the high-frequency calculator 81 calculates the margin subtraction frequency according to equation (23) as the high-frequency current frequency fh * Set as and the high-frequency current frequency fh after setting * The output is sent to the application method determination device 82.

number

[0129] On the other hand, in step S125, the high-frequency calculator 81 calculates the reference frequency fh1 to the high-frequency current frequency fh according to equation (24). * Set as and the high-frequency current frequency fh after setting * The output is sent to the application method determination device 82.

number

[0130] After processing in steps S110, S120, and S125, the process proceeds to step S130.

[0131] In step S130, the high-frequency calculator 81 calculates the high-frequency current frequency fh according to equation (25). * From high-frequency angular frequency ωh * The calculated high-frequency angular frequency ωh * The output is sent to the axis error calculator 30 and the phase generator 42.

number

[0132] As described above, the mechanical angular velocity command value ωm * The high-frequency current frequency fh is based on the cogging frequency fcog, which changes accordingly. * By changing this, it is possible to prevent fluctuation frequencies caused by cogging torque from being included in the pass-through frequency range of the common-mode mirror-phase current vector generator 301.

[0133] <Operation of the high-frequency calculator> Figure 15 shows an example of the operation of the high-frequency calculator of Embodiment 8 of this disclosure.

[0134] The high-frequency calculator 81 operates according to the flowchart shown in Figure 14, and as a result, the mechanical angular velocity command value ωm * and high-frequency current frequency fh * The relationship is shown in Figure 15. As can be seen from Figure 15, the motor control device 100d avoids current fluctuations that occur according to the rotational speed of the motor M by controlling the mechanical angular velocity command value ωm * Depending on the frequency of the high-frequency current fh * It changes.

[0135] The high-frequency calculator 81 may calculate the margin-added frequency by adding the first margin fmg1 to the first multiplication result obtained by multiplying the cogging frequency fcog by a first predetermined value α, and the margin-subtracted frequency by subtracting the second margin fmg2 from the second multiplication result obtained by multiplying the cogging frequency fcog by a second predetermined value β. The first predetermined value α is a positive integer greater than 0, and the second predetermined value β is the first predetermined value α plus 1. In Figures 14 and 15, the first predetermined value α is set to 1 as an example. In other words, the high-frequency calculator 81 calculates the high-frequency current frequency fh as a frequency that is greater than or equal to the frequency obtained by adding the first margin fmg1 to the first multiplication result obtained by multiplying the cogging frequency fcog by a first predetermined value α, and less than or equal to the frequency obtained by subtracting the second margin fmg2 from the second multiplication result obtained by multiplying the cogging frequency fcog by a second predetermined value β. * You can also set it as follows. By doing so, the high-frequency current frequency fh *Regardless of the value set, the high-frequency current frequency fh * This can be moved away from the fluctuation frequency.

[0136] As described above, the mechanical angular velocity command value ωm corresponds to the rotational speed of motor M. * Depending on the frequency of the high-frequency current fh * By changing this, the high-frequency current frequency can be set to a frequency that is sufficiently lower than the carrier frequency of the motor control device 100d and sufficiently higher than the frequency band of the drive voltage command value, thus keeping it away from fluctuation frequencies. This prevents the high-frequency current from being affected by current fluctuations. As a result, the rotor position can be estimated with high accuracy, thereby stabilizing the control of the motor M.

[0137] <Operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator> Figures 16 to 23 illustrate examples of operation of the application method determination device, noise reduction command generator, and high-frequency voltage command value generator of Embodiment 8 of this disclosure.

[0138] In the following, the conventional application method, that is, the application method in which the high-frequency voltage is continuously applied to the motor M without a change in the phase of the high-frequency voltage, will be referred to as "First Application Method NC1". In the following, the application method of Example 2, that is, the application method in which the high-frequency voltage is continuously applied to the motor M, and in addition, the high-frequency voltage is alternately applied to the motor as a first high-frequency voltage and a second high-frequency voltage which is in the opposite phase to the first high-frequency voltage, will be referred to as "Second Application Method NC2". In the following, the application method of Example 1, that is, the application method in which the high-frequency voltage is intermittently applied to the motor M, will be referred to as "Third Application Method NC3". In the following, the application method of Example 3, that is, the application method in which the high-frequency voltage is intermittently applied to the motor M, and in each section in which the high-frequency voltage is applied to the motor M, the high-frequency voltage is alternately applied to the motor M as a first high-frequency voltage and a second high-frequency voltage which is in the opposite phase to the first high-frequency voltage, will be referred to as "Fourth Application Method NC4".

[0139] Here, when the second application method NC2, the third application method NC3, or the fourth application method NC4 is used as the method for applying the high-frequency voltage, sidebands are generated. High-frequency current frequency fh * Based on this, the frequency of the sideband when the second application method NC2 is used is expressed by equation (26), the frequency of the sideband when the third application method NC3 is used is expressed by equation (27), and the frequency of the sideband when the fourth application method NC4 is used is expressed by equation (28). Here, X in equation (26) is the same as the number of intervals X in Example 2. Also, X and Y in equations (27) and (28) are the same as the number of applied intervals X and the number of unapplied intervals Y in Examples 1 and 3. In the following, among the sidebands expressed by equations (26), (27), and (28), the sideband with the higher frequency may be called the "upper sideband," and the sideband with the lower frequency may be called the "lower sideband." fh * ±fh * / 2X …(26) fh * ±fh * / (X+Y) …(27) fh * ±fh * / (2(X+Y)) …(28)

[0140] Furthermore, as the difference between the natural frequency SF of motor M and the frequency of the noise emitted by motor M due to the application of a high-frequency voltage decreases, the noise from motor M increases due to resonance. The smaller the difference between the natural frequency SF of motor M and the frequency of the noise emitted by motor M, the greater the noise from motor M increases due to resonance.

[0141] As an example, when the natural frequency SF of motor M is 250 Hz, the high-frequency current frequency fh * When it is 250Hz (natural frequency SF and high frequency current frequency fh * (If they match) and high-frequency current frequency fh * When it is 225Hz (natural frequency SF and high frequency current frequency fh *We will explain this separately for the case where the difference between the two is 25Hz. Here, for X and Y in equations (26), (27), and (28), we will explain them using the case where X=3 and Y=1, as in Examples 1 to 3.

[0142] <High frequency current frequency fh * If it's 250Hz > When the first application method NC1 is used as the method for applying high-frequency voltage, as shown in Figure 16, the frequency spectrum of the noise generated by the application of high-frequency voltage (hereinafter sometimes referred to as the "noise spectrum") appears only at 250 Hz. Furthermore, since the noise spectrum at 250 Hz coincides with the natural frequency SF, the noise at 250 Hz becomes louder.

[0143] Furthermore, when the second application method NC2 is used as the method for applying the high-frequency voltage, as shown in Figure 17, the time-averaged noise at 250 Hz generated by the application of the high-frequency voltage can be canceled out, while the sideband with the largest noise spectrum appears at 250 Hz ± 41.7 Hz.

[0144] Furthermore, when the third application method NC3 is used as the method for applying the high-frequency voltage, as shown in Figure 18, the time-averaged noise level at 250 Hz generated by the application of the high-frequency voltage can be reduced, while the sideband with the largest noise spectrum appears at 250 Hz ± 62.5 Hz. Also, since the noise spectrum at 250 Hz coincides with the natural frequency SF, the noise at 250 Hz becomes even louder.

[0145] Furthermore, when the fourth application method NC4 is used as the method for applying the high-frequency voltage, as shown in Figure 19, the time-averaged noise at 250 Hz generated by the application of the high-frequency voltage can be canceled out, while the sideband with the largest noise spectrum appears at 250 Hz ± 31.25 Hz.

[0146] Therefore, the natural frequency SF and the high-frequency current frequency fh *When these two conditions are met, the noise reduction effect can be enhanced by employing the second application method NC2 or the fourth application method NC4 as the method for applying high-frequency voltage. Note that the natural frequency SF and the high-frequency current frequency fh * Not only when they match, but also when the natural frequency SF and the high-frequency current frequency fh * Even when the difference is small (for example, within 10Hz), the noise reduction effect can be enhanced by adopting the second application method NC2 or the fourth application method NC4.

[0147] <High frequency current frequency fh * If it's 225Hz > When the first application method NC1 is used as the method for applying high-frequency voltage, the noise spectrum appears only at 225 Hz, as shown in Figure 20. Also, the high-frequency current frequency fh * Compared to the case where the frequency is 250Hz, the difference between the noise spectrum (225Hz) and the natural frequency SF (250Hz) is 25Hz, which reduces the noise.

[0148] Furthermore, when the second application method NC2 is used as the method for applying the high-frequency voltage, as shown in Figure 21, the time-averaged noise at 225 Hz generated by the application of the high-frequency voltage can be canceled out, while the sideband with the largest noise spectrum appears at 225 Hz ± 37.5 Hz.

[0149] Furthermore, when the third application method NC3 is used as the method for applying the high-frequency voltage, as shown in Figure 22, the time-averaged noise level at 225 Hz generated by the application of the high-frequency voltage can be reduced, while the sideband with the largest noise spectrum appears at 225 Hz ± 56.25 Hz.

[0150] Furthermore, when the fourth application method NC4 is used as the method for applying high-frequency voltage, as shown in Figure 23, while the time-averaged noise at 225 Hz generated by the application of high-frequency voltage can be canceled out, the sideband with the largest noise spectrum appears at 225 Hz ± 28.125 Hz. As a result, the noise increases because the difference between the frequency of the upper sideband (253.125 Hz) and the natural frequency SF (250 Hz) within the sideband frequency (225 Hz ± 28.125 Hz) is small at 3.125 Hz.

[0151] Therefore, when the difference between the frequency of the sidebands generated by the fourth application method NC4 and the natural frequency SF is small, the noise reduction effect can be enhanced by employing the second application method NC2 or the third application method NC3.

[0152] Also, high-frequency current frequency fh * Depending on the value, noise may not be noticeable even when the first application method NC1 is adopted. For example, if the natural frequency SF is 250 Hz, the high-frequency current frequency fh * This is the case when the frequency is sufficiently low at 200 Hz. In this case, it is preferable to adopt the first application method NC1, in which a high-frequency voltage is continuously applied to the motor M, in order to improve the accuracy of position estimation. Similarly, even when noise reduction is desired, if the difference between the frequency of the generated sidebands and the natural frequency SF does not decrease, it is preferable to adopt the second application method NC2, in which a high-frequency voltage is continuously applied to the motor M.

[0153] Therefore, the application method determination device 82 in Figure 13 uses the natural frequency SF of the motor M and the high-frequency current frequency fh *Based on the above, the method of applying high-frequency voltage is determined as follows. In the following, the margin frequency is denoted as "f_margin". f_margin is, for example, 10 Hz, and is set to a sufficiently large magnitude as the difference between the natural frequency SF and the sideband frequency so that resonance does not occur. In other words, if the difference between the sideband frequency and the natural frequency SF is greater than or equal to f_margin, it means that the sideband frequency does not overlap with the frequency band of the natural frequency SF, and the noise does not increase. Furthermore, the frequency of the upper sideband that occurs when the fourth application method NC4 is adopted is denoted as "f_upper_band_4", the frequency of the lower sideband that occurs when the fourth application method NC4 is adopted is denoted as "f_lower_band_4", the frequency of the upper sideband that occurs when the second application method NC2 is adopted is denoted as "f_upper_band_2", and the frequency of the lower sideband that occurs when the second application method NC2 is adopted is denoted as "f_lower_band_2". Furthermore, when the first application method NC1 is adopted, the high-frequency current frequency fh can be reduced to below the target value TG. * This is denoted as “f_noise_ok”. f_noise_ok is determined in advance through experiments, etc., as a frequency that is at least a predetermined frequency (e.g., 50 Hz) lower than the natural frequency. In addition, the application method adopted among the first application method NC1 to the fourth application method NC4 is sometimes called the “adopted application method”.

[0154] First, the application method determination device 82 determines the high-frequency current frequency fh * If the value is less than or equal to f_noise_ok, the system determines that the adopted application method is the first application method NC1 and outputs a signal_method set to “NOISE_CANCEL_OFF” to the noise reduction command generator 71. The noise reduction command generator 71, having received a signal_method set to “NOISE_CANCEL_OFF”, outputs a command to the high-frequency voltage command value generator 43 instructing it to generate a high-frequency voltage vector using the first application method NC1. In accordance with this command, the high-frequency voltage command value generator 43 generates a high-frequency voltage vector using the first application method NC1.

[0155] On the other hand, the application method determination device 82 determines the high-frequency current frequency fh * If the value is greater than f_noise_ok, it is determined that the adopted application method is either the second application method NC2, the third application method NC3, or the fourth application method NC4.

[0156] First, the application method determination device 82 calculates f_upper_band_2 according to equation (29) and then calculates f_lower_band_2 according to equation (30). Here, X in equation (29) is the same as the number of intervals X in Example 2. f_upper_band_2=fh * +fh * / 2X …(29) f_lower_band_2=fh * -fh * / 2X …(30)

[0157] Next, the application method determination unit 82 determines that the adopted application method is the second application method NC2 if f_upper_band_2 and f_lower_band_2 do not satisfy the conditions of equation (31), that is, if the sidebands when the second application method NC2 is used do not overlap with the frequency band of the natural frequency, and outputs the signal_method set to “NOISE_CANCEL_2” to the noise reduction command generator 71. The noise reduction command generator 71, which has received the signal_method set to “NOISE_CANCEL_2”, outputs the application method command signal_noise_method set as described in Example 2 to the high-frequency voltage command value generator 43, and according to this application method command, the high-frequency voltage command value generator 43 generates a high-frequency voltage vector using the second application method NC2. SF-f_margin≦f_upper_band_2≦SF+f_margin or SF-f_margin≦f_lower_band_2≦SF+f_margin …(31)

[0158] On the other hand, if f_upper_band_2 or f_lower_band_2 satisfies the conditions of equation (31), that is, if the sideband wave when the second application method NC2 is used overlaps with the frequency band of the natural frequency, the application method determination device 82 calculates f_upper_band_4 according to equation (32) and f_lower_band_4 according to equation (33). Here, X and Y in equations (32) and (33) are the same as the number of applied intervals X and the number of unapplied intervals Y in Example 3. f_upper_band_4=fh * +fh * / (2(X+Y)) …(32) f_lower_band_4=fh * -fh * / (2(X+Y)) …(33)

[0159] Next, the application method determination unit 82 determines that the adopted application method is the fourth application method NC4 if f_upper_band_4 and f_lower_band_4 do not satisfy the conditions of equation (34), that is, if the sidebands when the fourth application method NC4 is used do not overlap with the frequency band of the natural frequency, and outputs the signal_method set to “NOISE_CANCEL_4” to the noise reduction command generator 71. The noise reduction command generator 71, having received the signal_method set to “NOISE_CANCEL_4”, outputs the application method command signal_noise_method set as described in Example 3 to the high-frequency voltage command value generator 43, and according to this application method command, the high-frequency voltage command value generator 43 generates a high-frequency voltage vector using the fourth application method NC4. SF-f_margin≦f_upper_band_4≦SF+f_margin or SF-f_margin≦f_lower_band_4≦SF+f_margin …(34)

[0160] On the other hand, if f_upper_band_4 or f_lower_band_4 satisfies the conditions of equation (34), that is, if the sidebands when the fourth application method NC4 is used overlap with the frequency band of the natural frequency, the application method determined by the application method determination device 82 is the third application method NC3, and outputs the signal_method set to “NOISE_CANCEL_3” to the noise reduction command generator 71. The noise reduction command generator 71, having received the signal_method set to “NOISE_CANCEL_3”, outputs the application method command signal_noise_method, set as described in Example 1, to the high-frequency voltage command value generator 43, and according to this application method command, the high-frequency voltage command value generator 43 generates a high-frequency voltage vector using the third application method NC3.

[0161] In this way, the high-frequency voltage command value generator 43 determines the application method determination device 82 based on the natural frequency SF of the motor M and the high-frequency current frequency fh * Based on the method of applying high-frequency voltage determined from this, by applying high-frequency voltages with different waveforms, the high-frequency current frequency fh * Without changing the motor itself, the frequency of the sidebands generated by the high-frequency voltage can be changed. Therefore, an optimal application method can be adopted so that the sidebands do not overlap with the natural frequency band of the motor M, thereby suppressing motor noise.

[0162] As described above, the motor control device of this disclosure (motor control device 100d in the embodiment) includes an estimator (position estimator 32, PLL controller 31, and axis error calculator 30 in the embodiment) and an application unit (high frequency frequency calculator 81, application method determination unit 82, and high frequency voltage command value generator 43 in embodiment 8). The estimator estimates the position of the rotor of the motor based on the high frequency current generated when a high frequency voltage is applied to the motor (motor M in the embodiment). The application unit applies a high frequency voltage to the motor. The application unit can also change the frequency of the sidebands generated by the high frequency voltage.

[0163] For example, the application unit changes the frequency of the sidebands by applying high-frequency voltages of different waveforms. Alternatively, the application unit can apply high-frequency voltages of different waveforms by changing the application method used for applying the high-frequency voltage from among multiple application methods.

[0164] Furthermore, the application unit applies a high-frequency voltage to the motor using one of several application methods, in which the sidebands do not overlap with the motor's natural frequency band.

[0165] Furthermore, the multiple application methods include a first application method (first application method NC1 in Example 8) in which the high-frequency voltage is continuously applied to the motor without a change in phase. The application unit applies the high-frequency voltage to the motor using the first application method when the frequency of the high-frequency voltage is lower than or equal to a predetermined value (a predetermined frequency in Example 8) than the natural frequency of the motor.

[0166] Furthermore, the multiple application methods include a second application method (second application method NC2 in Example 8) in which a high-frequency voltage is continuously applied, and a first high-frequency voltage (first-phase high-frequency voltage in the embodiment) and a second high-frequency voltage (second-phase high-frequency voltage in the embodiment) which is in opposite phase to the first high-frequency voltage are alternately applied to the motor. The application unit applies the high-frequency voltage to the motor using the second application method when the frequency of the high-frequency voltage is not lower than a predetermined value (predetermined frequency in Example 8) than the natural frequency of the motor, and when the sidebands produced when the high-frequency voltage is applied to the motor using the second application method do not overlap with the frequency band of the motor's natural frequency.

[0167] Furthermore, the multiple application methods include a third application method (third application method NC3 in Example 8) in which a high-frequency voltage is intermittently applied to the motor.

[0168] Furthermore, the multiple application methods include a fourth application method (the fourth application method NC4 of Example 8) in which a high-frequency voltage is intermittently applied to the motor, and in each section in which the high-frequency voltage is applied to the motor, a first high-frequency voltage and a second high-frequency voltage that is in opposite phase to the first high-frequency voltage are alternately applied to the motor as the high-frequency voltage.

Explanation of Symbols

[0169] 100d Motor control device 28 Three-phase current calculator 30 Axis error calculator 301 In-phase mirror-phase current vector generator 302 Mirror-phase estimator 303, 305 Filter correction current estimator 31 PLL controller 32 Position estimator 41 High-frequency current subtractor 43 High-frequency voltage command value generator 61 Drive component removal filter 81 High-frequency frequency calculator 82 Application method determiner M Motor

Claims

1. An estimator that estimates the position of the rotor of the motor based on the high-frequency current generated when a high-frequency voltage is applied to the motor, The system includes an application unit that applies the aforementioned high-frequency voltage to the motor, The motor control device reduces the noise generated by the motor by changing the frequency of the sidebands, which are the spectrum of noise generated by the application of the high-frequency voltage, thereby generating the noise frequency spectrum as sidebands with a frequency different from the natural frequency of the motor.

2. The application unit changes the frequency of the sideband by applying a high-frequency voltage of a different waveform. The motor control device according to claim 1.

3. The application unit applies high-frequency voltages of different waveforms by changing the application method used for applying high-frequency voltage from among a plurality of application methods. The motor control device according to claim 2.

4. The application unit applies the high-frequency voltage to the motor using one of the plurality of application methods, wherein the sidebands do not overlap with the frequency band of the motor's natural frequency. The motor control device according to claim 3.

5. The plurality of application methods include a first application method in which the high-frequency voltage is continuously applied to the motor without changing its phase. The application unit applies the high-frequency voltage to the motor using the first application method when the frequency of the high-frequency voltage is lower than the natural frequency by a predetermined value or more. The motor control device according to claim 4.

6. The plurality of application methods include a second application method in which the high-frequency voltage is continuously applied, and the high-frequency voltage is alternately applied to the motor by a first high-frequency voltage and a second high-frequency voltage which is in opposite phase to the first high-frequency voltage. The motor control device according to claim 4.

7. The application unit applies the high-frequency voltage to the motor using the second application method when the frequency of the high-frequency voltage is not lower than a predetermined value below the natural frequency, and when the sidebands produced when the high-frequency voltage is applied to the motor using the second application method do not overlap with the frequency band of the natural frequency. The motor control device according to claim 6.

8. The plurality of application methods include a third application method in which the high-frequency voltage is intermittently applied to the motor. The motor control device according to claim 4.

9. The plurality of application methods include a fourth application method in which the high-frequency voltage is intermittently applied to the motor, and in each section in which the high-frequency voltage is applied to the motor, a first high-frequency voltage and a second high-frequency voltage having the opposite phase to the first high-frequency voltage are alternately applied to the motor as the high-frequency voltage. The motor control device according to claim 4.

Images