JP6372448B2 - Rotating machine control device - Google Patents

Description

  The present invention relates to a rotating machine control device that controls driving of a rotating machine using an output voltage of a pulse waveform.

  Conventionally, in controlling a rotating machine such as a motor, PWM control is widely used in which the fundamental wave voltage output from the controller is PWM modulated to control switching of the inverter. However, when the PWM frequency is high, the switching loss increases. On the other hand, if the PWM frequency is lowered in order to reduce the switching loss, the motor current ripple, particularly the d-axis current ripple with a small electric resistance value, increases, leading to an increase in iron loss.

  For example, Patent Document 1 discloses a technique for generating an output voltage pattern (hereinafter referred to as “pulse pattern”) having an optimal pulse waveform synchronized with the electrical angle of a motor, and controlling switching of the inverter using the pulse pattern. Is disclosed. When the pulse pattern is used, the number of times of switching in one electrical cycle can be reduced with respect to the PWM control, and the switching loss can be reduced. In the technique of Patent Document 1, an optimal pulse pattern is set so as to mainly reduce the power loss of the motor.

JP 2013-162660 A

  In the technique of Patent Document 1, “actual magnetic characteristic distortion” such as a motor distortion component, individual variation, and aging deterioration is not considered in generating a pulse pattern. Therefore, depending on the state of magnetic flux generated in the motor based on the set pulse pattern, the motor iron loss may increase. In particular, at the time of high rotation, the influence of motor iron loss appears remarkably.

  The present invention was created in view of the above points, and an object of the present invention is to consider an actual magnetic characteristic distortion of a rotating machine in a controller for a rotating machine that controls switching of an inverter using a pulse pattern. On the other hand, an object of the present invention is to provide a control device for a rotating machine that achieves both reduction of iron loss of the rotating machine and reduction of switching loss of the inverter.

The present invention relates to a control device for a rotating machine that controls switching of an inverter by a switching signal output from a modulator and controls energization of a multi-phase rotating machine having three or more phases. The modulator includes a magnetic flux prediction unit, a target magnetic flux setting unit, a magnetic flux error evaluation unit, and a pulse pattern generation unit.
The magnetic flux predicting unit integrates the voltage output from the inverter over a predetermined period, and calculates a predicted magnetic flux generated in the rotating machine by the switching operation of the inverter.
The target magnetic flux setting unit sets the target magnetic flux based on the actual magnetic characteristic distortion of the rotating machine.

  The magnetic flux error evaluation unit calculates an evaluation difference value based on a difference between the predicted magnetic flux and the target magnetic flux in a predetermined period. The pulse pattern generation unit generates a pulse pattern synchronized with the electrical angle of the rotating machine based on the voltage command value and closes the evaluation difference value to zero, and outputs a switching signal based on the generated pulse pattern. .

  In the rotating machine control device of the present invention, a pulse pattern is generated so that the evaluation difference value between the predicted magnetic flux and the target magnetic flux approaches zero, and switching of the inverter is controlled by a switching signal based on the pulse pattern. Here, the target magnetic flux is set based on the actual magnetic characteristic distortion of the rotating machine.

Thereby, the iron loss of a rotary machine can be reduced appropriately, considering the magnetic characteristic distortions, such as an actual magnet distortion. Therefore, it is possible to achieve both reduction of the iron loss of the rotating machine and reduction of the switching loss together with the effect of reducing the switching loss by the pulse pattern control.
The target magnetic flux setting unit can accurately reflect the actual magnetic characteristic distortion in the target magnetic flux by changing the target magnetic flux according to the rotational position of the rotating machine.

Further, it is preferable that the pulse pattern generation unit generates a pulse pattern in the electrical angle 360 [deg] section by inverting the line symmetry and the point symmetry with the section of the electrical angle 90 [deg] as one unit for one phase. .
Furthermore, it is preferable that the pulse pattern generation unit corrects the pattern based on a preset basic pattern so that the evaluation difference value calculated by the magnetic flux error evaluation unit approaches zero.

The schematic block diagram of the control apparatus of the rotary machine by one Embodiment of this invention. The block diagram of the modulator of FIG. (A) The figure explaining (alpha) (beta) coordinate system, (b) The figure which shows the magnetic flux vector of a prediction magnetic flux and a target magnetic flux. Diagram explaining the difference in magnetic flux vector The time chart explaining the setting of the pulse pattern based on magnetic flux error evaluation. The figure explaining the definition of a pulse pattern. (A), (b) PWM pattern (prior art), (c) Magnetic flux locus by the same pattern. (A), (b) Basic pattern, (c) Magnetic flux trajectory by the same pattern. (A), (b) 1st correction pattern, (c) Magnetic flux locus by the same pattern. (A), (b) 2nd correction pattern, (c) Magnetic flux locus by the same pattern. (A), (b) 3rd correction pattern, (c) Magnetic flux locus | trajectory by the pattern. (A), (b) The 4th correction pattern, (c) The figure of the magnetic flux locus by the pattern. The figure which shows each evaluation difference value ((a) phase difference, (b) amplitude difference, (c) difference in phase change, (d) difference in amplitude change) between the predicted magnetic flux and the target magnetic flux by the PWM pattern (prior art) . The figure which shows the evaluation difference value of the prediction magnetic flux by a basic pattern, and a target magnetic flux. The figure which shows the evaluation difference value of the prediction magnetic flux and target magnetic flux by a 1st correction pattern. The figure which shows the evaluation difference value of the prediction magnetic flux by a 2nd correction pattern, and a target magnetic flux. The figure which shows the evaluation difference value of the prediction magnetic flux and target magnetic flux by a 3rd correction pattern. The figure which shows the evaluation difference value of the prediction magnetic flux and target magnetic flux by a 4th correction pattern. The figure explaining the motor current ripple and iron loss by PWM control of a prior art.

Hereinafter, a control device for a rotating machine according to an embodiment of the present invention will be described based on the drawings.
(One embodiment)
A schematic configuration and operation of a control device for a rotating machine according to an embodiment will be described with reference to FIGS. In the present embodiment, the three-phase motor corresponds to a “multi-phase rotating machine”, and the motor control device corresponds to a “rotating machine control device”. As shown in FIG. 1, the motor control device 10 outputs a switching signal to the inverter 4 connected to the motor 5 to control switching. Thereby, the electric power supplied from the inverter 4 to the motor 5 is controlled.

  The inverter 4 includes three-phase upper and lower arm switching elements 41 to 46 that are bridge-connected. Specifically, switching elements 41, 42, and 43 are U-arm, V-phase, and W-phase upper arm switching elements, respectively, and switching elements 44, 45, and 46 are respectively under the U-phase, V-phase, and W-phase. This is an arm switching element. As the switching elements 41 to 46, for example, an IGBT or the like is used. The inverter 4 converts the input DC power into three-phase AC power by the operation of each switching element 41 to 46.

The motor 5 is, for example, a permanent magnet type synchronous three-phase AC motor. In this embodiment, as the motor 5, for example, a motor generator that is a power source of a hybrid vehicle or an electric vehicle is assumed. A position sensor 55 such as a resolver for detecting the rotor position (electrical angle) θe is provided in the vicinity of the rotor of the motor 5.
A current sensor 6 that detects three-phase currents Iu, Iv, and Iw is provided in the current path from the inverter 4 to the motor 5. The current sensor 6 may detect a two-phase current, and the other one-phase current may be calculated according to Kirchhoff's law. Alternatively, a technique for estimating the current of the other two phases based on the detected current value of one phase, or a technique of estimating the current based on the detected DC current value may be employed.

The motor control device 10 includes a dq conversion unit 11, a current subtractor 12, a current controller 13, an inverse dq conversion unit 14, a modulator 15, a differentiator 16, and the like as a configuration of current feedback control.
The dq conversion unit 11 receives the phase current detection value from the current sensor 6. The dq conversion unit 11 uses the electrical angle θe acquired from the position sensor 55 to dq convert the three-phase current detection values Iu, Iv, and Iw into dq-axis current detection values Id and Iq, and feeds back to the current subtractor 12. .

The current subtracter 12 subtracts the dq-axis current detection values Id and Iq fed back from the dq converter 11 from the dq-axis current command values Id ^* and Iq ^* to calculate dq-axis current deviations ΔId and ΔIq.
The current controller 13 calculates the dq axis voltage command values Vd ^* and Vq ^* by PI control calculation or the like so that the dq axis current deviations ΔId and ΔIq converge to zero.
The inverse dq converter 14 converts the dq axis voltage command values Vd ^* and Vq ^* into the three-phase voltage command values Vu ^* , Vv ^* and Vw ^* using the electrical angle θe.

Modulator 15 receives voltage command values Vu ^* , Vv ^* , Vw ^* , electrical angle θe, and electrical angular velocity ω obtained by differentiating electrical angle θe with time by differentiator 16. Here, the electrical angular velocity ω may be read as the motor rotation speed. Based on this information, the modulator 15 switches the switching signals UU, VU, WU corresponding to the switching elements 41, 42, 43 of the three-phase upper arm and the switching elements 44, 45, three-phase of the lower arm. The switching signals UL, VL, WL corresponding to 46 are generated and output to the inverter 4.
Further, the modulator 15 acquires an induced voltage e when the inverter 4 is stopped.

As shown in FIG. 2, the modulator 15 includes a pulse pattern generation unit 20, a magnetic flux prediction unit 31, a target magnetic flux setting unit 32, and a magnetic flux error evaluation unit 33.
First, the motor control device 10 of the present embodiment controls switching of the inverter 4 using a pulse pattern synchronized with the electrical angle of the motor 5 instead of PWM control widely used in motor control technology. And When the pulse pattern is used, the number of times of switching in one electrical cycle can be reduced with respect to the PWM control, and the switching loss can be reduced. In particular, since the motor generator of a hybrid vehicle is required to be small and have high output, performance in a high rotation overmodulation region is important. Therefore, it is effective to reduce switching loss using a pulse pattern.

  The pulse pattern generation unit 20 preferably includes a basic pattern setting unit 21, a pattern correction unit 23, and storage units 22 and 24. The basic pattern setting unit 21 sets a “basic pattern” based on the voltage command value, the electrical angle θe, and the angular velocity ω (number of rotations). In this case, a plurality of pulse patterns are stored in advance in the storage unit 22 in the form of a map or the like, and stored in the storage unit 22 when the basic pattern setting unit 21 acquires the voltage command value, the electrical angle θe, and the electrical angular velocity ω. An optimum pattern may be selected from the plurality of pulse patterns. Alternatively, the modulation factor and the voltage phase may be calculated from the voltage command value and the inverter DC voltage, and the pattern may be selected from the modulation factor and the voltage phase.

  Conventionally, as an index for selecting the “optimum pattern”, a technique for minimizing power loss and torque ripple is known. In the present embodiment, however, attention is paid to the actual magnetic flux of the motor 5. Therefore, the pattern correction unit 23 acquires an “evaluation difference value”, which will be described later, from the magnetic flux error evaluation unit 33, acquires the electrical angle θe, and corrects the basic pattern so that the evaluation difference value approaches zero. Is generated. Specifically, the pattern correction unit 23 generates a pulse pattern so that the peak value, the integrated value, or the average value of the evaluation difference values in a predetermined period approaches zero.

The correction pattern generated by the pattern correction unit 23 may be stored in the storage unit 24 in the form of a map or the like representing the relationship with each parameter, in the same manner as the basic pattern is stored in the storage unit 22. The pattern correction unit 23 may further re-correct the corrected pattern read from the storage unit 24.
The pulse pattern generation unit 20 outputs a switching signal (UU, VU, WU, UL, VL, WL) based on the correction pattern thus obtained to the inverter 4.

  The switching signal generated by the pulse pattern generation unit 20 is acquired by the magnetic flux prediction unit 31. The magnetic flux predicting unit 31 integrates the voltage output from the inverter 4 over a predetermined period based on the switching signal, and predicts the actual magnetic flux generated in each phase of the motor 5 by the switching operation of the inverter 4. The magnetic flux predicted by the magnetic flux predicting unit 31 based on the pulse pattern is referred to as “predicted magnetic flux Φest”.

When the voltage is v, the resistance is R, and the current is I, the predicted magnetic flux Φest is calculated by Expression (1.1). The magnetic flux unit [Wb] has a relationship of [Wb] = [V · s].

However, since the RI term is sufficiently smaller than the voltage v in the high rotation region, the equation (1.2) ignoring the RI term can be used.

Since the iron loss becomes a problem particularly in the high rotation range, the formula (1.2) is practically used. The magnetic flux predicting unit 31 may switch the calculation so that, for example, the formula (1.1) is used when the rotational speed is less than the predetermined rotational speed, and the formula (1.2) is used when the rotational speed is the predetermined rotational speed or higher.
Further, it is assumed that the control of the present embodiment is performed in a state where the rotation speed is stable. Therefore, the electrical angular velocity ω (= dθe / dt) may be regarded as a constant, and the integration interval of the equation (1.2) may be converted from the time t to the electrical angle θe as in the equation (1.3).

  The target magnetic flux setting unit 32 sets the target magnetic flux Φtgt for the predicted magnetic flux Φest based on the actual magnetic characteristic distortion of the motor 5. The target magnetic flux Φtgt includes a component such as a magnetic flux distortion of a permanent magnet or an inductance distortion of a coil as “real magnetic characteristic distortion”. Etc.

  For example, the target magnetic flux setting unit 32 acquires the induced voltage e of the motor 5 when the inverter 4 is stopped, and sets the target magnetic flux Φtgt based on the magnet magnetic flux waveform detected from the induced voltage e. Alternatively, the target magnetic flux setting unit 32 may internally store distortion information inspected for each individual when the motor 5 is manufactured, and may set the target magnetic flux Φtgt based on the distortion information.

The magnetic flux error evaluation unit 33 calculates an evaluation difference value used for evaluating a difference between the predicted magnetic flux Φest and the target magnetic flux Φtgt in a predetermined period.
Here, FIG. 3 is referred with respect to fixed coordinates (αβ coordinates) at which the magnetic flux error is evaluated in the present embodiment. FIG. 3A shows αβ coordinates based on the U phase. Conversion from the UVW three-phase axis to the α-axis and β-axis two-phase axes is expressed by Expression (2). The α axis indicates the component in the U phase direction, and the β axis indicates the component in the direction orthogonal to the U phase. Further, the phase θe of the electrical angle is defined counterclockwise with respect to the α axis.

  On the αβ coordinate axis, the magnetic flux vector is expressed as shown in FIG. Assuming that the magnetic characteristic distortion of the motor 5 is uniform regardless of the position, the broken line target magnetic flux Φtgt is ideally circular. However, the target magnetic flux Φtgt is a distorted circle when the magnetic characteristic distortion varies depending on the position or between the three phases. Thus, the target magnetic flux setting unit 32 changes the target magnetic flux Φtgt according to the rotational position of the motor 5.

  The predicted magnetic flux Φest calculated based on the pulse pattern appears in rotational symmetry with 60 [deg] as one unit. In general, the predicted magnetic flux Φest is a polygonal line that straddles the target magnetic flux Φtgt. Therefore, the portion where the predicted magnetic flux Φest exceeds the target magnetic flux Φtgt and the portion where the predicted magnetic flux Φest falls below the target magnetic flux Φtgt appear alternately depending on the phase.

Subsequently, an error between the predicted magnetic flux Φest and the target magnetic flux Φtgt will be schematically described with reference to FIG. The phase of the predicted magnetic flux vector Φest (t) at time t is expressed as θest (t), and the phase of the target magnetic flux vector Φtgt (t) is expressed as θtgt (t).
The magnetic flux phase difference Δθ is calculated by the equation (3.1).
Δθ (t) = θest (t) −θtgt (t) (3.1)
Further, the magnetic flux amplitude difference ΔΦa is calculated by the equation (3.2).
ΔΦa (t) = | Φest (t) | − | Φtgt (t) | (3.2)

Further, the difference in magnetic flux phase change is calculated by equation (3.3), and the difference in magnetic flux amplitude change is calculated by equation (3.4).
d (Δθ (t)) / dt = d (θest (t) −θtgt (t)) / dt
... (3.3)
d (ΔΦa (t)) / dt = d (| Φest (t) | − | Φtgt (t) |) / dt
... (3.4)
In addition, you may set the positive / negative of the sign in Formula (3.1)-(3.4) suitably.

The magnetic flux phase difference, the magnetic flux amplitude difference, the magnetic flux phase change difference, or the magnetic flux amplitude change difference can all be used for the error evaluation between the predicted magnetic flux Φest and the target magnetic flux Φtgt in the magnetic flux error evaluation unit 33. Value. In the present specification, these difference values are referred to as “evaluation difference values”.
As described above, the magnetic flux error evaluation unit 33 calculates the evaluation difference value on the fixed coordinate (αβ coordinate) axis. Then, the pattern correction unit 23 of the pulse pattern generation unit 20 generates a correction pattern so that the evaluation difference value approaches zero.

Next, the operation of the modulator 15 will be described with reference to the time chart of FIG.
When the magnetic flux predicting unit 31 calculates the predicted magnetic flux Φest and the magnetic flux error evaluating unit 33 calculates the evaluation difference value between the predicted magnetic flux Φest and the target magnetic flux Φtgt in the predetermined period, the pattern correcting unit 23 corrects the correction pattern based on the evaluation difference value. Is generated. The correction pattern is reflected in the subsequent control cycle. For example, the pattern T1 is generated based on the evaluation difference value in a certain predetermined period T1, and the next pattern T2 is generated based on the evaluation difference value in the next predetermined period T2.
Assuming that this control is performed under the condition that the rotation speed is constant, the electrical angle section of the phase axis is converted into a period of the time axis. In FIG. 5, for example, a period corresponding to one period of electrical angle is illustrated as a predetermined period on the time axis.

Here, as shown in FIG. 3B, the evaluation difference value of the three-phase motor periodically varies with the electrical angle 60 [deg] as one unit.
Therefore, it is preferable that the predetermined period for evaluating the magnetic flux error between the predicted magnetic flux Φest and the target magnetic flux Φtgt is a period corresponding to the electrical angle 60 [deg] or more of the motor 5. Stable evaluation is possible by evaluating the magnetic flux error over a period of 60 electrical degrees or more.
If the predetermined period is set to a period that is an integral multiple of the control period, the evaluation result can be efficiently reflected in the pulse pattern calculation in the next control period.

Next, the definition and setting procedure of the pulse pattern will be described with reference to FIG.
As shown in FIG. 6A, the pulse pattern of one phase (for example, U phase) among the three phases is set by a model having a section of an electrical angle of 90 [deg] as one unit.
First, referring to the electrical angle in the upper stage (b) with respect to the horizontal axis of FIG. 6 (a), in the example of FIG. 6 (a), two off periods and two times in the section of electrical angle 0 to 90 [deg]. Is set to the ON period. The pulse width in the first off period from the electrical angle 0 [deg] is δ1. In addition, the center position of the second off period is θc1, and half the width of the second off period is the pulse width δ2. If the center position θc1 and the pulse widths δ1 and δ2 are determined, the pulse pattern in the section of the electrical angle 0 to 90 [deg] is determined.

Then, the pulse pattern with the electrical angle of 90 to 180 [deg] as shown by the broken line is set by inverting the pulse pattern with the electrical angle of 0 to 90 [deg] around the electrical angle of 90 [deg]. .
Also, the pulse pattern of electrical angles 180 to 360 [deg] is set by inverting the pulse pattern of electrical angles 0 to 180 [deg] around the electrical angle 180 [deg] as a point symmetry. “Point symmetry” means to invert line symmetry and to invert the on and off sides of the pulse.

As described above, a pulse pattern of one electric cycle for one phase (for example, U phase) is determined. The other two phases (V phase and W phase) are set by shifting the pulse pattern of one phase by an electrical angle of ± 120 [deg]. For example, the pulse pattern in the section of the electrical angle 360 [deg] of the U phase and the V phase is expressed as shown in FIG.
This pulse pattern means ON / OFF of the switching elements 41, 42, and 43 of the upper arms of each phase. If the dead time is ignored, the pulse pattern of the switching elements 44, 45, and 46 in the lower arm of each phase is complementary to the pulse pattern of the switching elements 41, 42, and 43 in the same phase, that is, a vertically inverted form. Become.

  When each phase switching element 41-46 operates according to a switching signal based on each phase pulse pattern, inverter 4 outputs a voltage. FIG. 6B shows a line voltage pulse pattern of the U-V phase based on the U-phase and V-phase pulse patterns. The line voltage pulse pattern has a “half-wave symmetrical” relationship in which the voltage in the electrical angle range of 330 to 150 [deg] and the voltage in the electrical angle range of 150 to 330 [deg] are inverted.

  By the way, in defining the pulse pattern, the reference of the electrical angle may be appropriately changed. For example, with respect to the horizontal axis in FIG. 6A, the solid line section in FIG. 6A may be defined as a section of 270 to 360 [deg] using the electrical angle shown in parentheses in the lower row (c). In this case, the pulse pattern of electrical angles 270 to 360 [deg] is inverted symmetrically about the electrical angle 360 [deg] (= 0 [deg]), and the pulse of electrical angle 0 to 90 [deg] is obtained. Set the pattern. Further, the pulse pattern of electrical angles 90 to 270 [deg] is set by inverting the pulse pattern of electrical angles 270 to 360 and 0 to 90 [deg] around the electrical angle 90 [deg] in a point-symmetric manner.

Then, the pulse pattern in the electrical angle 360 [deg] section is represented by FIG. This pulse pattern corresponds to a sinusoidal fundamental voltage indicated by a broken line. In the following FIGS. 7 to 12, the pulse pattern displayed in this electrical angle section is described.
The “pulse number n of pulse pattern” means the number of pulses of each phase pulse pattern between electrical angles 360 [deg] or the number of pulses of the line voltage pulse pattern between electrical angles 180 [deg]. Say. In the example of FIG. 6, “n = 7”. Hereinafter, unless otherwise specified, “pulse pattern” means “each phase pulse pattern”.

Next, examples of specific pulse patterns and magnetic flux trajectories between the predicted magnetic flux Φest and the target magnetic flux Φtgt based on those pulse patterns are shown in FIGS.
Specifically, each of FIGS. (A) and (b) shows a pulse pattern in the range of 90 [deg] and 360 [deg]. Also, in each figure (c), the magnetic flux trajectories of the predicted magnetic flux Φest and the target magnetic flux Φtgt are shown by αβ coordinates according to FIG. 3 (b).

First, FIG. 7 shows a pattern when the number of pulses is 7 in the conventional PWM control.
In the locus of the predicted magnetic flux Φest in the αβ coordinate, a zigzag waveform appears every electrical angle 60 [deg]. In the peaks and valleys of the zigzag waveform, the amplitudes of the predicted magnetic flux Φest and the target magnetic flux Φtgt are different. In the PWM pattern, the deviation of the magnetic flux amplitude is relatively large.

Next, the basic pattern of this embodiment is shown in FIG. This basic pattern is set so that the number of pulses is the same as that of the PWM pattern, and the effective values are equal. The basic pattern is set so that the pulse intervals near the electrical angles 90 [deg] and 270 [deg] are uniform. That is, if the pulse width of the first off period is δ1, the pulse width of the first on period is also δ1, and δ2, which is a half of the pulse width of the second off period, is 0. It corresponds to 5 times. Therefore, Formula (4) is materialized.
θc1 = 2 × δ1 + δ2 = 2.5 × δ2 (4)
Compared with the PWM pattern, the basic pattern has a smaller amplitude divergence from the zigzag waveform target magnetic flux Φtgt in the locus of the predicted magnetic flux Φest.

Further, four kinds of correction patterns obtained by changing a part of the basic pattern are shown in FIGS. Among them, the number of pulses of the first to third correction patterns is 7 as in the basic pattern.
The first correction pattern shown in FIG. 9 is obtained by changing the pulse width from δ2 to δ2 ^* with respect to the basic pattern.
The second correction pattern shown in FIG. 10 is obtained by changing the pulse position, that is, the center position θc1 to θc1 ^* with respect to the basic pattern.

The third correction pattern shown in FIG. 11 is obtained by changing both the pulse width and the pulse position to δ2 ^** and θc1 ^** so that the effective values are equal to those of the basic pattern.
In these first to third correction patterns, the number of bends of the zigzag waveform in the locus of the predicted magnetic flux Φest is the same as that of the basic pattern, and the shape and size of the zigzag waveform is slightly different from the basic pattern.

  The fourth correction pattern shown in FIG. 12 is obtained by increasing the number of pulses from 7 to 11 with respect to the basic pattern. The number of bends of the zigzag waveform in the locus of the predicted magnetic flux Φest increases from 7 to 11 with respect to the basic pattern, and the size of each zigzag is smaller than the basic pattern as a whole.

  Next, the evaluation of the difference between the predicted magnetic flux Φest and the target magnetic flux Φtgt by the pulse pattern will be described with reference to FIGS. In each figure, the horizontal axis represents an electrical angle, and the vertical axis represents four evaluation difference values: (a) magnetic flux phase difference, (b) magnetic flux amplitude difference, (c) magnetic flux phase change difference, (d) magnetic flux Indicates the difference in amplitude change. A specific unit is not shown in each evaluation difference value, and relative evaluation is performed using the scale [div] on the vertical axis as an index.

  Regarding the dimension of each evaluation difference value, (a) the magnetic flux phase difference is an angular dimension such as [deg] or [rad], and (b) the magnetic flux amplitude difference is a magnetic flux dimension such as [Wb]. Further, (c) the difference in magnetic flux phase change is the angular velocity dimension of [deg / s] or [rad / s], and (d) the difference in magnetic flux amplitude change is a voltage of [Wb / s] = [V]. Dimension.

  In the conventional PWM pattern shown in FIG. 13, (b) the width from the maximum peak to the minimum peak of the magnetic flux amplitude difference (hereinafter, “width between peaks”) is about 3.0 [div]. Further, (c) the maximum peak of the difference in magnetic flux phase change is about 1.5 [div], and (d) the maximum peak of the difference in magnetic flux amplitude change is about 3.0 [div], which is relatively large.

  Here, FIG. 19 is referred to regarding the current ripple in the PWM control. In general, in PWM control, switching loss increases when the PWM frequency is high. On the other hand, when the PWM frequency is lowered in order to reduce the switching loss, the motor current ripple, particularly the d-axis current ripple having a small electric resistance value, increases. And the iron loss in a coil iron core increases.

  In addition, when switching is controlled using a pulse pattern disclosed in Patent Document 1 (Japanese Patent Laid-Open No. 2013-162660) or the like for PWM control, the number of times of switching in one electrical cycle can be reduced and switching loss can be reduced. it can. However, the motor iron loss depends on a motor distortion component such as a magnetic flux distortion of a permanent magnet or an inductance distortion of a coil, a magnetic characteristic distortion due to individual variation, aging deterioration, and the like. Therefore, even if the pulse pattern is set without considering the motor distortion, the motor iron loss cannot be effectively reduced.

  As shown in FIG. 13, that the evaluation difference value between the predicted magnetic flux Φest and the target magnetic flux Φtgt is relatively large means that the motor iron loss is likely to increase. Therefore, the present embodiment aims to reduce the motor iron loss by generating a pulse pattern so that the evaluation difference value approaches zero and controlling the switching of the inverter 4 based on the pulse pattern. .

  Here, as a method of bringing the evaluation difference value close to zero, the peak value of the evaluation difference value in a predetermined period may be close to zero. Or you may make it the integral value or average value of the evaluation difference value in a predetermined period approach zero. The use of an integral value or an average value is less susceptible to noise than the use of a peak value. For the average value, for example, the weighted average may be calculated so that the more recent values are weighted.

  As shown in FIG. 14, in the basic pattern of this embodiment, (b) the width between the peaks of the magnetic flux amplitude difference is about 2.4 [div], which is smaller than the width of the PWM pattern. Further, (c) the maximum peak of the difference in magnetic flux phase change is about 1.2 [div], and (d) the maximum peak of the difference in magnetic flux amplitude change is about 2.5 [div], both of which are compared with the PWM pattern. Small. That is, the peak value of the evaluation difference value approaches zero. Therefore, by controlling the switching of the inverter 4 based on the basic pattern of the present embodiment, the motor iron loss can be reduced as compared with the conventional technique using the PWM pattern.

As shown in FIGS. 15, 16, and 17, all of the first, second, and third correction patterns are compared with the PWM pattern, (b) the width between peaks of the magnetic flux amplitude difference, and (c) the magnetic flux phase. The maximum peak of the difference in change and (d) the maximum peak of the difference in magnetic flux amplitude change are small. In particular, when comparing the width between the peaks of the magnetic flux amplitude difference (b), the first correction pattern is about 2.0 [div], the second correction pattern is about 2.0 [div], and the third correction pattern is about 1. 8 [div]. That is, it is smaller than about 2.4 [div] of the basic pattern.
Note that (c) the maximum peak of the difference in magnetic flux phase change and (d) the maximum peak of the difference in magnetic flux amplitude change of the first, second, and third correction patterns are similar to the basic pattern.

  As described above, in the first and second correction patterns in which the pulse width or the pulse position is changed with respect to the basic pattern, and in the third correction pattern in which the pulse width and the pulse position are changed so that the effective values are equal to the basic pattern. The peak value of the evaluation difference value can be further reduced. Therefore, by controlling the switching of the inverter 4 based on the first to third correction patterns, the motor iron loss can be further reduced as compared with the case where the basic pattern is used.

In addition, as shown in FIG. 18, even in the fourth correction pattern in which the number of pulses is increased with respect to the basic pattern, compared with the PWM pattern, (b) the width between peaks of the magnetic flux amplitude difference, and (c) the change in magnetic flux phase. The maximum peak of the difference and (d) the maximum peak of the difference in magnetic flux amplitude change are small. The width between the peaks of the (b) magnetic flux amplitude difference of the fourth correction pattern is about 1.8 [div], which is the minimum level along with the third correction pattern.
However, in the fourth correction pattern, the number of pulses is increased with respect to the basic pattern, and the switching loss tends to increase. Therefore, it is preferable to determine whether or not the fourth correction pattern is appropriate after comprehensively considering the switching loss and the motor iron loss.

(effect)
The effect of the motor control device 10 of the present embodiment will be described.
(1) In this embodiment, a pulse pattern is generated so that the evaluation difference value between the predicted magnetic flux Φest and the target magnetic flux Φtgt is close to zero, and switching of the inverter 4 is controlled by a switching signal based on the pulse pattern. Here, the target magnetic flux Φtgt is set based on the actual magnetic characteristic distortion of the motor 5.

As a result, the motor iron loss can be appropriately reduced while considering the magnetic characteristic distortion such as the actual magnet distortion. Therefore, the reduction of the motor iron loss and the reduction of the switching loss can be achieved together with the effect of reducing the switching loss by the pulse pattern control.
Further, the target magnetic flux setting unit 32 can accurately reflect the actual magnetic characteristic distortion on the target magnetic flux Φtgt by changing the target magnetic flux Φtgt according to the rotational position of the motor 5.

  (2) In the present embodiment, by the method shown in FIG. 6, the section of the electrical angle of 90 [deg] is set as one unit for one phase, and the pulse in the section of electrical angle of 360 [deg] is inverted by line symmetry and point symmetry. Generate a pattern. With such a pattern generation method, the amount of pattern storage can be reduced, and the calculation load for generating a new pattern can be reduced.

  (3) The pulse pattern generation unit 20 of the present embodiment is configured so that the evaluation difference value calculated by the magnetic flux error evaluation unit 33 approaches zero based on the basic pattern previously set by the basic pattern setting unit 21. The correction unit 23 generates a correction pattern. In addition, the basic pattern and the correction pattern are stored in the storage units 22 and 24 together with the parameters used for calculating the pattern.

  In this way, by selecting a basic pattern stored in advance in the storage unit 22 and performing a correction calculation based on the basic pattern to generate a correction pattern, the calculation load can be reduced and the calculation time can be shortened. it can. Furthermore, it is possible to learn efficiently by storing the correction pattern in the storage unit 24 anew.

(Other embodiments)
(A) In the above-described embodiment, the electrical angle ± 120 [deg] is shifted from the other two-phase pulse pattern with respect to the single-phase pulse pattern, and the electrical angle is 60 [deg] or more as a predetermined period for evaluating the magnetic flux error. The period corresponding to is set because the number of phases of the motor is three phases. On the other hand, the number of phases of the multiphase rotating machine to which the present invention is applied may be four or more. In this case, the electrical angle is appropriately set according to the number of phases.

(A) In contrast to the pulse pattern generation unit 20 of the above embodiment that includes the basic pattern setting unit 21 and the pattern correction unit 23, the pulse pattern generation unit of other embodiments may not include the basic pattern setting unit. That is, the pulse pattern for making the evaluation difference value close to zero may be directly calculated without setting the basic pattern once. In addition, the pulse pattern generation unit does not include a storage unit, and an optimal pulse pattern may be calculated from the beginning each time. Further, the storage unit 24 of the pattern correction unit 23 may not be provided, and the pattern correction unit 23 may correct the storage unit 22 of the basic pattern unit 21.
In short, the pulse pattern generation unit of the present invention may output a switching signal based on the pulse pattern generated so that the evaluation difference value approaches zero to the inverter.

  (C) The rotating machine to be controlled in the present invention is not limited to a permanent magnet type synchronous motor such as IPMSM and SPMSM, but all rotating machines that have a coil and can cause iron loss, such as an induction motor and a switched reluctance motor. included. Furthermore, the rotating machine is not limited to a motor generator or the like used as a power source for a vehicle, but may be used for auxiliary equipment of a vehicle, a train other than the vehicle, an elevator, a general machine, or the like.

(D) In the present invention, the method for controlling the energization of the motor is not limited to the current feedback control as shown in FIG. 1, but may be torque feedback control. Further, voltage open control that directly generates a voltage command value from a torque command or a rotation speed command without performing current feedback may be used.
As mentioned above, this invention is not limited to the said embodiment at all, In the range which does not deviate from the meaning of invention, it can implement with a various form.

10: Motor control device (control device for rotating machine),
15 ... modulator,
20 ... pulse pattern generation unit,
31 ... Magnetic flux prediction part,
32 ... Target magnetic flux setting section,
33 ... Magnetic flux error evaluation unit,
4 ... Inverter
5: Motor (motor generator, rotating machine).

Claims (15)

A control device for a rotating machine that controls switching of an inverter (4) by a switching signal output from a modulator (15) and controls energization of a multi-phase rotating machine (5) of three or more phases,
The modulator is
A magnetic flux predicting unit (31) that integrates a voltage output from the inverter over a predetermined period, and calculates a predicted magnetic flux generated in the rotating machine by the switching operation of the inverter;
A target magnetic flux setting unit (32) for setting a target magnetic flux based on the actual magnetic characteristic distortion of the rotating machine;
A magnetic flux error evaluation unit (33) for calculating an evaluation difference value used for evaluating a difference between the predicted magnetic flux and the target magnetic flux in the predetermined period;
Pulse pattern generation based on a voltage command value and generating a pulse pattern synchronized with the electrical angle of the rotating machine so that the evaluation difference value approaches zero and outputting the switching signal based on the generated pulse pattern Part (20);
A control device for a rotating machine.   2. The target magnetic flux includes a component of a magnetic flux distortion of a permanent magnet constituting the rotating machine or an inductance distortion of a coil as an actual magnetic characteristic distortion of the rotating machine. Rotating machine control device.   The said target magnetic flux setting part sets the said target magnetic flux based on the magnet magnetic flux waveform detected from the induced voltage of the said rotary machine at the time of a stop of the said inverter, The control apparatus of the rotary machine of Claim 2 characterized by the above-mentioned. .   The said target magnetic flux setting part changes the said target magnetic flux according to the rotation position of the said rotary machine, The control apparatus of the rotary machine of Claim 2 or 3 characterized by the above-mentioned. The magnetic flux error evaluation unit, as the evaluation difference value,
5. The rotating machine according to claim 1, wherein a phase difference, an amplitude difference, a phase change difference, or an amplitude change difference between the predicted magnetic flux and the target magnetic flux is calculated. Control device. The pulse pattern generation unit
6. The control device for a rotating machine according to claim 5, wherein a pulse pattern is generated so that a peak value of the evaluation difference value in the predetermined period approaches zero. The pulse pattern generation unit
6. The control apparatus for a rotating machine according to claim 5, wherein the pulse pattern is generated so that an integral value or an average value of the evaluation difference values in the predetermined period approaches zero. The multi-phase rotating machine is a three-phase rotating machine,
The control apparatus for a rotating machine according to any one of claims 1 to 7, wherein the predetermined period is a period corresponding to an electrical angle of 60 [deg] or more of the rotating machine.   The control apparatus for a rotating machine according to any one of claims 1 to 8, wherein the predetermined period is a period that is an integral multiple of a control cycle. The multi-phase rotating machine is a three-phase rotating machine,
The pulse pattern generation unit
About one-phase pulse pattern among three phases,
Set the section of electrical angle 0-90 [deg] as one unit,
The pulse pattern of electrical angle 90-180 [deg] is set by reversing the pulse pattern of electrical angle 0-90 [deg] around the electrical angle 90 [deg] in a line symmetry,
A pulse pattern of electrical angles 180 to 360 [deg] is set by inverting the pulse pattern of electrical angles 0 to 180 [deg] around the electrical angle 180 [deg] in a point-symmetric manner,
The control device for a rotating machine according to any one of claims 1 to 9, wherein the two-phase pulse pattern is set by shifting the one-phase pulse pattern by an electrical angle of ± 120 [deg]. The pulse pattern generation unit
A basic pattern setting unit (21) for setting a basic pattern which is a basic pulse pattern;
A pattern correction unit (23) that generates a correction pattern in which the evaluation difference value is corrected to approach zero with respect to the basic pattern;
The control device for a rotating machine according to any one of claims 1 to 10, wherein:   The control device for a rotating machine according to claim 11, wherein the pattern correction unit changes the pulse width of the basic pattern to generate the correction pattern.   The control apparatus for a rotating machine according to claim 11, wherein the pattern correction unit changes the pulse position of the basic pattern to generate the correction pattern.   The rotating machine control device according to claim 11, wherein the pattern correction unit generates the correction pattern so that an effective value is equal to the basic pattern.   The control device for a rotating machine according to claim 11, wherein the pattern correction unit changes the number of pulses of the basic pattern to generate the correction pattern.

Images