JP5312195B2 - DC brushless motor control device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate a phase difference estimate value at which a waveform is kept suitable for stably and accurately estimating a rotor angle, regardless of the state of occurrence of magnetic saturation. <P>SOLUTION: Based on the observed value of change in current with a high-frequency periodical signal superimposed on the driving voltage of an armature, the following values are computed (STEP6): a phase difference basic reference value &Theta;Eb which varies according to the double angle of the phase difference &theta;e between the estimate value and the actual value of a rotor angle and causes the waveform of the variation to vary depending on the state of occurrence of magnetic saturation in a motor; and an inductance reference value L0 depending on the average inductance of the motor. A phase difference estimate value &Theta;E is computed by linearly coupling the phase difference basic reference value &Theta;Eb and the inductance reference value L0 (STEP7). The estimate value of the rotor angle is updated according to this phase difference estimate value &Theta;E (STEP12). <P>COPYRIGHT: (C)2011,JPO&amp;INPIT

Description

  The present invention relates to a control device for a DC brushless motor.

  When controlling the operation of a DC brushless motor equipped with a saliency rotor, without using a magnetic pole position detection sensor such as a resolver (so-called sensorless), the rotational angle position of the rotor magnetic pole (rotation at an electrical angle) A technique for estimating a rotor angle indicating (angular position) has been proposed by the present applicant (see, for example, Patent Document 1).

  This technology observes the current change of the armature for each predetermined period with the high-frequency periodic signal with a predetermined pattern superimposed on the drive voltage applied to the armature of the motor, and estimates the rotor angle from the observed value. A sine reference value as a value according to the sine function value sin (2θe) of the double angle 2θe of the phase difference θe between the actual value and the inductance reference value as a value according to the average inductance of the motor is calculated. Then, a value obtained by normalizing the sine reference value with the inductance reference value is obtained as a phase difference estimated value that changes according to the phase difference between the estimated value of the rotor angle and the actual value (true value), and this phase difference estimated value Accordingly, the estimated value of the rotor angle is sequentially updated.

JP 2007-143275 A

  By the way, in the rotor angle estimation technique in Patent Document 1, in order to stably converge the estimated value of the rotor angle to the actual value (in order to converge the phase difference θe to “0”), The waveform of the phase difference estimation value for the double angle 2θe is always a substantially constant waveform, and the waveform of the phase difference estimation value when the value of the phase difference θe is positive and the value of the phase difference θe are negative. Ideally, the waveform of the phase difference estimation value when the value of is a symmetrical waveform with opposite polarities. For example, it is preferable that the waveform of the phase difference estimation value is always a substantially constant sine waveform.

  However, in a state where a current is applied to the armature of the motor so that an output torque other than “0” is generated in the motor (a state where the magnitude of the current is relatively large), in general, the magnetic saturation of the motor does not occur. Likely to happen. When the magnetic saturation occurs, the waveform of the phase difference estimated value with respect to the phase difference θe becomes a waveform that is collapsed from the sinusoidal waveform. As a result, the amplitude of the waveform of the phase difference estimated value becomes unbalanced between the positive side and the negative side, and the range of the phase difference θe where the phase difference estimated value becomes a positive value and the phase difference estimated value become a negative value. One of the ranges of the phase difference θe may be too narrow.

  In such a case, the estimated value of the rotor angle can be updated so that the phase difference θe converges stably to “0” in a situation where the estimated value of the phase difference is a positive value or a negative value. The range of the phase difference θe (hereinafter sometimes referred to as a stable convergence range) is narrow. Therefore, the estimated value of the rotor angle may diverge from the actual value.

  Here, in the technique found in Patent Document 1, an appropriate offset value is added to the phase difference estimated value. By doing in this way, it is possible to prevent the stable convergence range from becoming too narrow. However, in this case, the convergence value of the estimated value of the rotor angle deviates from the actual value, and the error of the estimated value of the rotor angle increases. For this reason, when the estimated value of the rotor angle is used to control the output torque of the motor, there is a tendency that the actual output torque becomes lower than the target torque.

  The present invention has been made in view of such a background, and generates a phase difference estimation value that is maintained in a waveform suitable for stable and accurate estimation of the rotor angle regardless of the state of occurrence of magnetic saturation. Therefore, an object of the present invention is to provide a control device for a DC brushless motor capable of estimating the rotor angle stably and accurately regardless of the state of occurrence of magnetic saturation.

  In order to achieve the above object, the DC brushless motor control apparatus according to the present invention includes a current detection means for detecting a phase current flowing in the armature of the DC brushless motor, and a rotor angle estimation means for estimating the rotor angle of the motor. Then, by executing a predetermined feedback calculation process using the detected value of the phase current by the current detecting means and the estimated value of the rotor angle by the rotor angle estimating means, the detected value of the phase current is set to a target current. Dq basic voltage determining means for determining a basic value of a d-axis voltage and a basic value of a q-axis voltage that define a drive voltage to be applied to the armature for convergence, and at least a basic value of the d-axis voltage And a control device for a DC brushless motor that applies a drive voltage defined by a basic value of a q-axis voltage to the armature to control energization of the armature. High-frequency superimposing means for applying to the armature a high-frequency superimposed drive voltage obtained by superimposing a high-frequency periodic signal on a drive voltage defined by the basic value of the axial voltage and the basic value of the q-axis voltage; In the applied state, changes in the d-axis current and the q-axis current corresponding to the change in the phase current of the armature are observed based on the detected value of the phase current and the estimated value of the rotor angle. Based on the observed value of the change in current, the value changes according to the double angle of the phase difference between the estimated value of the rotor angle and the actual value of the rotor angle, and the waveform of the change with respect to the double angle is Basic reference value calculation means for calculating a phase difference basic reference value that changes depending on the state of occurrence of magnetic saturation, and for calculating an inductance reference value as a value that depends on the average inductance of the motor; By linearly combining the difference basic reference value and the inductance reference value, the change waveform with respect to the phase difference is suppressed from changing depending on the state of occurrence of magnetic saturation of the motor, compared to the phase difference basic reference value. Phase difference estimated value calculating means for calculating the phase difference estimated value, wherein the rotor angle estimating means updates the estimated value of the rotor angle according to the phase difference estimated value.

  In the present invention, the d-axis current and the q-axis current are dq coordinates in which one of the magnetic flux direction of the field pole of the rotor of the DC brushless motor and the direction perpendicular thereto is the d-axis direction and the other is the q-axis direction. When the current flowing through the armature of the motor is expressed in the system, it means the current flowing through the d-axis armature (armature on the d-axis) and the q-axis armature (armature on the q-axis).

  Here, when the phase difference basic reference value calculated based on the observed value of the change in the d-axis current and the q-axis current and the inductance reference value are linearly combined using appropriate weighting factors applied to each, It is possible to keep the waveform of the change of the linear combination value with respect to the phase difference in a substantially constant waveform even when the state of occurrence of magnetic saturation of the motor changes. That is, by the linear combination, it is possible to obtain a reference value that changes in accordance with the phase difference while suppressing dependency on the state of occurrence of magnetic saturation of the motor.

  Therefore, in the first invention, at least the phase difference basic reference value and the inductance reference value are linearly combined to change the waveform of the change with respect to the phase difference depending on the state of occurrence of magnetic saturation of the motor. Phase difference estimated value calculating means for calculating a phase difference estimated value that is suppressed more than the phase difference basic reference value is provided.

  This makes it possible to calculate a phase difference estimated value that is not easily affected by magnetic saturation. And the said rotor angle estimation means updates the estimated value of the said rotor angle according to this phase difference estimated value.

  Therefore, according to the first aspect of the present invention, it is possible to generate a phase difference estimation value that maintains a waveform suitable for stable and accurate estimation of the rotor angle regardless of the state of magnetic saturation of the motor. As a result, the rotor angle can be estimated stably and accurately regardless of the state of occurrence of magnetic saturation.

  In the first invention, the phase difference estimated value calculation means includes at least one of a weighting coefficient applied to the phase difference basic reference value in the linear combination and a weighting coefficient applied to the inductance reference value. It is preferable to use a weighting factor that is set so that the value changes depending on the output torque of the motor (second invention).

  That is, as the motor output torque increases, the magnitude of the phase current that flows through the armature of the motor increases, and therefore magnetic saturation of the motor tends to occur. Therefore, the state of occurrence of magnetic saturation of the motor has a dependency on the output torque of the motor. Therefore, in the second invention, the weighting factor is set to change depending on the output torque of the motor.

  As a result, the linear combination can be performed using an optimum weighting factor in suppressing the change in the waveform of the phase difference estimation value (the waveform of the change with respect to the phase difference) according to the state of occurrence of magnetic saturation of the motor. It becomes. As a result, it is possible to generate a phase difference estimation value such that the waveform of the change with respect to the phase difference is maintained in a substantially constant shape regardless of the state of occurrence of magnetic saturation of the motor.

  In the first and second aspects of the invention, the phase difference estimated value calculation means is a value obtained by linearly combining the phase difference basic reference value and the inductance reference value depending on the output torque of the motor. It is preferable to calculate the phase difference estimated value by adding a correction term set so as to change (third invention).

  That is, the waveform of the change with respect to the phase difference of the value obtained by linearly combining the phase difference basic reference value and the inductance reference value is suppressed from changing its shape according to the state of magnetic saturation of the motor. Even in this case, the entire waveform may be offset depending on the state of occurrence of magnetic saturation of the motor. Therefore, in the third aspect of the invention, a correction term set so that the value changes depending on the output torque of the motor is added to a value obtained by linearly combining the phase difference basic reference value and the inductance reference value. The phase difference estimation value is calculated.

  As a result, the waveform shape of the phase difference estimated value change with respect to the phase difference changes according to the state of occurrence of magnetic saturation, and the entire offset of the waveform is suppressed from occurring according to the state of occurrence of magnetic saturation. It is possible to calculate the estimated phase difference.

  As a result, the rotor angle can be estimated more accurately and stably regardless of the state of magnetic saturation of the motor. As a result, the phase current flowing through the armature of the motor is controlled to an appropriate current for generating the target output torque in the motor without being affected by the state of magnetic saturation of the motor. It becomes possible.

  In the present invention, it is preferable to calculate the phase difference basic reference value and the inductance reference value by the following configuration, for example.

  That is, n high-frequency periodic signals to be superimposed by the high-frequency superimposing means on a drive voltage defined by the basic value of the d-axis voltage and the basic value of the q-axis voltage (hereinafter sometimes referred to as a basic drive voltage). A pattern is set in advance in which the sum of the voltage values of the periodic signals in a predetermined period including (n: an integer of 2 or more) continuous control cycles is “0”.

  The dq basic voltage determining means determines a basic value of the d-axis voltage and a basic value of the q-axis voltage so as to keep the basic driving voltage constant during the predetermined period. Further, the reference value calculation means calculates the first-order difference of the d-axis current and the first-order difference of the q-axis current in the period between each control cycle and the next control cycle within the predetermined period as the phase current. The phase difference basic reference value and the inductance reference value are calculated by a predetermined calculation process using the observed value and the estimated value of the rotor angle.

  In this case, for example, the value of L0 on the left side calculated from the observation value of the first-order difference of the d-axis current and the observation value of the first-order difference of the q-axis current by the calculation of the right side of Equation (1) below. Can be used as the inductance reference value.

Further, for example, the value of Vsdq calculated by the calculation of the right side of Expression (1) (the value of the first component of the vertical vector of the left side of Expression (1)) can be used as the phase difference basic reference value. . Alternatively, the value of Vsdq on the left side and the value of L0 are calculated by the calculation of the right side of Equation (1), and the value Vsdq / L0 obtained by dividing (normalizing) the calculated value of Vsdq by the value of L0 is You may make it set as a phase difference basic reference value. Alternatively, the value of Vsdq and the value of Vcdq on the left side are calculated by the calculation of the right side of Equation (1), and the calculated value of Vsdq is calculated as the square value of the value of Vsdq and the square value of the value of Vcdq. A value (= Vsdq / √ (Vsdq ² + Vcdq ² )) obtained by dividing by the square root of the sum (= √ (Vsdq ² + Vcdq ² )) may be set as the phase difference basic reference value.

  Cdq in equation (1) is a matrix defined by the following equations (2) and (3), and dIdq (i + j) (j = 1, 2,..., N) is expressed by the following equation (4). A vertical vector defined. In this case, Vhd (i + j) (j = 1, 2,..., N) in the expression (3) is expressed as d when the periodic signal is applied to the d-axis armature in the jth control cycle within the predetermined period. The axis voltage (d-axis voltage component when the voltage of the periodic signal is expressed in the dq coordinate system), Vhq (i + j) (j = 1, 2,..., N) is the jth value within the predetermined period. The periodic signal is a q-axis voltage applied to the q-axis armature in a control cycle (a q-axis voltage component when the voltage of the periodic signal is expressed in the dq coordinate system). In addition, dId (i + j) (j = 1, 2,..., N) in Expression (4) is a d-axis current in a period between the j-th control cycle and the next control cycle within the predetermined period. , DIq (i + j) (j = 1, 2,..., N) is the q axis in the period between the jth control cycle and the next control cycle within the predetermined period. It is an observed value of the first-order difference of current.

  Further, the first-order difference dId (i + j) of the d-axis current is obtained by replacing the d-axis current Id (i + j) in the i + j-th control cycle with the d-axis current Id (i + j +) in the next control cycle. It is calculated as a value obtained by dividing the value (= Id (i + j + 1) −Id (i + j)) subtracted from 1) by the time ΔT of one control cycle period. The same applies to the first-order difference dIq (i + j) of the q-axis current.

  In the expressions (1) to (4), i means the number of the control cycle immediately before the control cycle in which the predetermined period of the periodic signal starts.

Supplementally, as the phase difference basic reference value, in addition to Vsdq, Vsdq / L0, Vsdq / √ (Vsdq ² + Vcdq ² ) exemplified above, for example, each value is multiplied by a proportional constant of a predetermined value (≠ 0). This value may be used again as the phase difference basic reference value. The same applies to the inductance reference value.

  Here, although details will be described later, L0 calculated by the calculation of the right side of the equation (1) is Ld as the average inductance of the d-axis armature of the motor and Lq as the average inductance of the q-axis armature. These Ld and Lq are values having a relationship (L0 = ((1 / Ld) + (1 / Lq)) / 2) shown in the following formula (7). Therefore, L0 exemplified above as an inductance reference value (or an inductance reference value obtained by multiplying L0 by a proportional constant of a predetermined value) has a meaning as a value depending on the average inductance of the motor.

  Further, Vsdq calculated by the calculation of the right side of the equation (1) from the observed values dId and dIq of the first-order difference of the d-axis current and the q-axis current, and the actual value and the estimated value of the rotor angle within the predetermined period. In principle, a relationship of Vsdq = L1 · sin (2θe) is established between the phase difference θe (= actual value−estimated value) and the phase difference θe (= actual value−estimated value). To do. In principle, between Vcdq calculated by the calculation of the right side of the formula (1) and the phase difference θe within the predetermined period (at least in a state where magnetic saturation of the motor has not occurred), The relationship Vcdq = L1 · cos (2θe) is established.

  Note that L1 is a value having the relationship (L1 = − ((1 / Ld) − (1 / Lq)) / 2) expressed by the equation (7) described later with respect to the average inductances Ld and Lq. It is.

Therefore, the phase difference basic reference values Vsdq, Vsdq / L0, Vsdq / √ (Vsdq ² + Vcdq ² ) (or the phase difference basic reference values obtained by multiplying these values by a predetermined proportional constant) are Specifically, it has a meaning as a value (a value proportional to sin2θe) corresponding to a sine function value (sin2θe) of a double angle of the phase difference θe. However, these phase difference basic reference values generally have a waveform in which the waveform of the change with respect to the double angle of the phase difference θe collapses from a sinusoidal waveform when the magnetic saturation of the motor occurs.

  Then, the phase difference basic reference value and the inductance reference value exemplified above are linearly combined using appropriate weighting factors (weighting factors respectively applied to the phase difference basic reference value and the inductance reference value), so that the motor Regardless of the state of occurrence of magnetic saturation, it is possible to calculate a phase difference estimated value such that the waveform of the change with respect to the phase difference θe is kept constant.

The block diagram which shows the whole structure of the control apparatus in 1st Embodiment of this invention. FIGS. 2A and 2B are diagrams illustrating examples of voltage vectors in a dq coordinate system of a periodic signal superimposed on an armature drive voltage in the first embodiment. 3A and 3B are graphs illustrating waveforms of the phase difference basic reference value ΘEb in the first embodiment. The graph which illustrates the waveform of inductance reference value L0 in a 1st embodiment. The flowchart which shows the process of the principal part of the control apparatus in 1st Embodiment shown in FIG. FIGS. 6A to 6C are graphs illustrating data tables for determining the weighting factors α and β and the correction term γ used for calculating the phase difference estimation value ΘE in the first embodiment. FIGS. 7A to 7C are graphs illustrating waveforms of the phase difference estimated value ΘE calculated in the first embodiment. FIG. 8A is a graph illustrating a change in the actual value θ of the rotor angle and a change in the estimated value θ ^ in the first embodiment, and FIG. 8B is a diagram of the motor at the time of estimating the rotor angle in the present embodiment. The graph which illustrates the change of output torque. FIG. 9A is a graph illustrating a change in the actual value θ of the rotor angle and a change in the estimated value θ ^ in the conventional rotor angle estimation method, and FIG. 9B is a conventional rotor angle estimation method. The graph which illustrates the change of the output torque of the motor at the time of estimation of the rotor angle of. FIGS. 10A to 10C illustrate data tables for determining the weighting factors α and β and the correction term γ, respectively, used for calculating the phase difference estimation value ΘE in the second embodiment of the present invention. To graph. FIGS. 11A to 11C are graphs illustrating waveforms of the phase difference estimated value ΘE calculated in the second embodiment.

  A first embodiment of the present invention will be described below with reference to FIGS.

  In FIG. 1, 1 is a DC brushless motor, and 10 is a control device that controls the operation of a DC brushless motor 1 (hereinafter simply referred to as a motor 1).

  Although not shown in detail, the DC brushless motor 1 (hereinafter simply referred to as the motor 1) is an IPM (Interior Permanent Magnet) type synchronous motor in which a plurality of permanent magnets are embedded in the rotor. The rotor has saliency. A stator (not shown) arranged around the rotor of the motor 1 is mounted with a plurality of armatures, for example, U-phase, V-phase, and W-phase armatures 2u, 2v, and 2w. Yes. In the following description, when it is not necessary to distinguish the armatures 2u, 2v, and 2w of the respective phases, they are simply referred to as the armature 2.

  The control device 10 includes an arithmetic processing unit 11 including a CPU, a RAM, a ROM, and the like, and an armature of any two phases among the three phases, for example, a U-phase armature 2u and a W-phase armature 2w. It is composed of current sensors 12u and 12w that detect energization currents (phase currents) and a power drive unit 13 (hereinafter referred to as PDU 13) constituted by an inverter circuit having a plurality of switch elements (not shown).

  In the present embodiment, the current detection means in the present invention is realized by the current sensors 12u and 12w.

  The sum of the phase currents of the armatures 2u, 2v, and 2w of the U phase, V phase, and W phase is “0”. Therefore, if the phase current of the armature 2 of any two of the three phases is detected, the remaining one-phase phase current can also be specified from the detected phase current values. For this reason, in this embodiment, two current sensors 12u and 12w are provided. However, the phase currents of the three phases may be detected by different current sensors.

  The PDU 13 is connected to a DC power source (not shown) as a power source of the motor 1 and is connected to the armature 2 of each phase. Furthermore, phase voltage command values Vu_c, Vv_c, and Vw_c sequentially calculated by the arithmetic processing unit 11 as will be described later are input to the PDU 13 as control commands that define the operation. The phase voltage command values Vu_c, Vv_c, and Vw_c are command values of drive voltages (phase voltages) to be applied to the U-phase armature 2u, the V-phase armature 2v, and the W-phase armature 2w, respectively. Then, the PDU 13 performs the ON / OFF operation of the switch element of the inverter circuit by the PWM method according to the input phase voltage command values Vu_c, Vv_c, and Vw_c. Thereby, the PDU 13 applies the drive voltages of the phase voltage command values Vu_c, Vv_c, and Vw_c from the DC power source to the armature 2 of each phase, and energizes the armature 2 of each phase.

  The arithmetic processing unit 11 receives the U-phase phase current detection value Iu_s indicated by the output of the current sensor 12u and the W-phase phase current detection value Iw_s indicated by the output of the current sensor 12w.

  The arithmetic processing unit 11 has a function of feedback control to a target current with respect to the phase current of the armature 2 of each phase. In this case, in this embodiment, in order to perform the calculation process of the feedback control, for example, the arithmetic processing unit 11 sets the magnetic flux direction of the field pole of the rotor of the motor 1 (the field magnetic pole by the permanent magnet) to the q axis direction. A dq coordinate system, which is a rotating coordinate system (coordinate system rotating at the electrical angular velocity of the rotor) with the direction orthogonal to the axial direction as the d-axis direction, is used. The arithmetic processing unit 11 converts the motor 1 into an equivalent circuit including an armature on the q axis (q axis armature) and an armature on the d axis (d axis armature). That is, the arithmetic processing unit 11 expresses the current and voltage of the entire armature 2 of the motor 1 as a vector quantity on the dq coordinate system (a vector quantity composed of a d-axis direction component and a q-axis direction component). The phase current of the armature 2 of each phase of the motor 1 is feedback-controlled by vector control arithmetic processing.

  In the present embodiment, the magnetic flux direction of the field pole of the rotor of the motor 1 is defined as the q-axis direction, but the magnetic flux direction may be defined as the d-axis direction.

In the following description, the d-axis current as the energization current of the d-axis armature and the q-axis current as the energization current of the q-axis armature in the dq coordinate system are denoted by reference numerals Id and Iq, respectively. Further, a d-axis voltage as a voltage of the d-axis armature and a q-axis voltage as a voltage of the q-axis armature are denoted by reference signs Vd and Vq, respectively. Further, a vector (Id, Iq) ^t (subscript t means transposition) composed of the d-axis current Id and the q-axis current Iq is obtained from the dq current vector Idq, the d-axis voltage Vd, and the q-axis voltage Vq. The constructed vector (Vd, Vq) ^t may be referred to as a dq voltage vector Vdq.

  The arithmetic processing unit 11 is so-called sensorless (without using a sensor for detecting the rotor angle such as a resolver), and the rotor angle indicating the magnetic pole position of the rotor of the motor 1 (rotational angle position at the electrical angle of the rotor). It has the function to estimate. This function is basically used as a drive voltage (drive voltage for feedback control) to be applied to the armature 2 of each phase in order to converge the phase current of each phase to the target current in order to estimate the rotor angle. This includes a function of applying a high frequency superimposed drive voltage obtained by superimposing a high frequency periodic signal (voltage signal) to the drive voltage to the armature 2 of each phase. Then, the arithmetic processing unit 11 converts the estimated value θ ^ of the calculated rotor angle between the phase current or phase voltage of the motor 1 and the current or voltage in the dq coordinate system in the vector control arithmetic processing. Used for.

  These functions are realized by processing that the arithmetic processing unit 11 sequentially executes in a predetermined control cycle. Details of the processing of the arithmetic processing unit 11 will be described below.

  The arithmetic processing unit 11 calculates the d-axis current detection value Id_s as the detection value of the d-axis current Id from the input phase current detection values Iu_s and Iw_s, and the q-axis current detection value Iq_s as the detection value of the q-axis current Iq. 3 (in other words, the detected value of the dq current vector Idq) is provided.

The three-phase-dq converter 21 is a vector (Iu_s, Iw_s) ^t having phase current detection values Iu_s, Iw_s as components, or phase current detection values Iu_s, Iw_s and a V-phase phase current detection value Iv_s (however, A detected value (Id_s, Iq_s) ^t of the dq current vector Idq is calculated by multiplying a vector (Iu_s, Iv_s, Iw_s) ^t having a component of Iv_s = − (Iu_s + Iw_s)) by a predetermined transformation matrix. As is well known, the transformation matrix is a transformation matrix in which the component values of each row and each column are determined according to the rotor angle. In this case, the three-phase-dq conversion unit 21 uses the estimated value θ ^ of the rotor angle calculated by the rotor angle estimation unit 30 described later as the value of the rotor angle for determining each component value of the conversion matrix. .

  Further, the arithmetic processing unit 11 sets a set of a d-axis current command value Id_c that is a target value of the d-axis current Id and a q-axis current command value Iq_c that is a target value of the q-axis current Iq (in other words, a dq current vector). Current command value determination unit 22 that sequentially determines (command value of Idq), and d-axis current deviation ΔId (= Id_c_−Id_s) that is a deviation between the d-axis current command value Id_c and the detected d-axis current value Id_s d An axis current deviation calculating unit 23, a q axis current deviation calculating unit 24 for calculating a q axis current deviation ΔIq (= Iq_c_−Iq_s) which is a deviation between the q axis current command value Iq_c and the q axis current detected value Iq_s, and d A set of a d-axis voltage FB command value Vfbd and a q-axis voltage FB command value Vfbq as a target value of the dq voltage vector Vdq required for converging the shaft current deviation ΔId and the q-axis current deviation ΔIq to “0” is fed back. Current FB control unit 25 calculated using a control law, and motor A set of a d-axis voltage command value Vd_c and a q-axis voltage command value Vq_c as a target value of a dq voltage vector Vdq that defines a set of drive voltages to be actually applied to the armatures 2 of each phase of the phase voltage command And a dq-3 phase converter 26 for converting into a set of values Vu_c, Vv_c, and Vw_c.

In this case, in this embodiment, the torque command value Tr_c, which is the target value of the output torque of the motor 1, is sequentially input from the outside to the current command value determination unit 22 of the arithmetic processing unit 11. Then, the current command value determination unit 22 determines, based on the input torque command value Tr_c, a preset data table (a relationship between Tr_c and Iq_c and a relationship between Tr_c and Id_c, respectively). The command value (Id_c, Iq_c) ^t of the dq current vector Idq is determined based on the table. The data table is such that the command value (Id_c, Iq_c) ^{t of the} dq current vector Idq determined thereby becomes an appropriate command value for causing the motor 1 to efficiently generate the output torque of the torque command value Tr_c. Is set. In this case, in this embodiment, for example, the reluctance torque due to the saliency of the motor 1 is efficiently generated, and the output torque of the torque command value Tr_c is reduced while reducing the current flowing through the armature 2 of the motor 1 as much as possible. The command value (Id_c, Iq_c) ^t of the dq current vector Idq is determined so that it can be generated at 1.

  Further, the current FB control unit 25 uses the PI law (proportional / integral law) as a feedback control law from the d-axis current deviation ΔId and the q-axis current deviation ΔIq, for example, and uses the d-axis voltage FB command values Vfbd and q The shaft voltage FB command value Vfbq is calculated respectively. Specifically, the current FB control unit 25 integrates a proportional term obtained by multiplying the d-axis current deviation ΔId by a proportional gain of a predetermined value and an integral value of ΔId (cumulative addition value for each control cycle) of a predetermined value. A feedback manipulated variable (basic value of the d-axis voltage FB command value Vfbd) for converging ΔId to “0” is calculated by adding together the integral term obtained by multiplying. Similarly, the current FB control unit 25 multiplies a proportional term obtained by multiplying the q-axis current deviation ΔIq by a proportional gain of a predetermined value, and an integral value of ΔIq (cumulative addition value for each control cycle) by an integral gain of a predetermined value. In addition, the feedback manipulated variable (basic value of the q-axis voltage FB command value Vfbq) for converging ΔIq to “0” is calculated.

  The feedback control law is not limited to the PI law, and other feedback control law such as a proportional law may be used.

  Then, the current FB control unit 25 performs the basic of Vfbd by the non-interacting process for canceling the influence of the speed electromotive force that interferes between the d-axis and the q-axis (the influence on the q-axis current and the d-axis current). The d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq are determined by correcting the value and the basic value of Vfbq.

  In this non-interacting process, the d-axis voltage correction amount Vdx, which is the correction amount of the d-axis side voltage command value for canceling the influence of the speed electromotive force, is used as the q-axis current detection value Iq_s and the rotor described later. The angle estimation unit 30 calculates the estimated value ω ^ of the electrical angular velocity of the rotor of the motor 1 that is calculated together with the estimated value θ ^ of the rotor angle based on a predetermined arithmetic expression or map. Similarly, the q-axis voltage correction amount Vqx, which is the correction amount of the voltage command value on the q-axis side, is calculated from the d-axis current detection value Id_s and the estimated value ω ^ of the electrical angular velocity of the rotor of the motor 1 by a predetermined arithmetic expression. Or it calculates based on a map. A value obtained by adding the d-axis voltage correction amount Vdx to the basic value of Vfbd is calculated as a d-axis voltage FB command value Vfbd. Further, a value obtained by adding the q-axis voltage correction amount Vqx to the basic value of Vfbq is calculated as a q-axis voltage FB command value Vfbq.

  In a situation where the influence of the speed electromotive force that interferes between the d-axis and the q-axis is sufficiently small, the decoupling processing is omitted, and the basic value of Vfbd and the basic value of Vfbq are May be determined as the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq, respectively.

In the following description, a dq voltage vector (Vfbd, Vfbq) ^t composed of a set of the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq may be referred to as a dq voltage FB command vector Vfbdq.

In the present embodiment, the three-phase-dq conversion unit 21, the d-axis current deviation calculation unit 23, the q-axis current deviation calculation unit 24, and the current FB control unit 25 realize the dq basic voltage determination unit in the present invention. The In this case, the three-phase current corresponding to the command value (Id_c, Iq_c) ^t of the dq current vector Idq corresponds to the target current. Further, the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq correspond to the basic value of the d-axis voltage and the basic value of the q-axis voltage, respectively.

The dq-3 phase converter 26 applies a predetermined conversion matrix to a command value (Vd_c, Vq_c) ^t of a dq voltage vector Vdq composed of a set of an input d-axis voltage command value Vd_c and a q-axis voltage command value Vq_c. By multiplying, a vector (Vu_c, Vv_c, Vw_c) ^t composed of a set of phase voltage command values Vu_c, Vv_c, Vw_c is calculated. Similar to the conversion matrix used in the conversion process of the three-phase-dq conversion unit 21, the conversion matrix is a conversion matrix (three-phase-dq conversion) in which the component values of each row and each column are determined according to the rotor angle. A conversion matrix for performing a conversion process reverse to the conversion process in the unit 21). In this case, the dq-3 phase conversion unit 26 uses the rotor angle estimation value θ calculated by the rotor angle estimation unit 30 described later as a rotor angle value for determining each component value of the conversion matrix in the processing. Use ^.

  Here, in the feedback control of the phase current of the armature 2 of each phase of the motor 1 so that the d-axis current deviation ΔId and the q-axis current deviation ΔIq converge to “0”, the current FB control unit 25 The dq voltage FB command vector Vfbdq consisting of the set of the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq calculated as described above is input to the dq-3 phase conversion unit 26 as the command value of the dq voltage vector Vdq. Thus, the phase voltage command values Vu_c, Vv_c, and Vw_c may be converted into a set and the Vu_c, Vv_c, and Vw_c may be input to the PDU 13. At this time, drive voltages of phase voltage command values Vu_c, Vv_c, and Vw_c obtained by converting the dq voltage FB command vector Vfbdq are applied to the armature 2 of each phase. As a result, the phase current flowing through the armature 2 of each phase of the motor 1 is feedback controlled to the target phase current (phase current defined by the set of the d-axis current command value Id_c and the q-axis current command value Iq_c). Will be.

  On the other hand, the rotor angle estimation method in the present embodiment estimates the rotor angle using the fact that the inductance of the rotor of the motor 1 changes according to the rotor angle, so that the rotational speed of the rotor is “0” or The rotor angle can be estimated even at a low speed. For this purpose, the arithmetic processing unit 11 additionally superimposes a high-frequency periodic signal (voltage signal) on the drive voltage applied to the armature 2 of each phase.

  In this case, in this embodiment, on the dq coordinate system (before the conversion process of the dq-3 phase conversion unit 26), a set of the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq has a high frequency. The periodic signal is superimposed.

  Therefore, the arithmetic processing unit 11 represents the vector amount (the voltage vector of the periodic signal viewed in the dq coordinate system) when the voltage of the periodic signal superimposed on the driving voltage of the armature 2 is expressed as a vector amount on the dq coordinate system. ) Of a d-axis direction component Vhd (hereinafter referred to as a d-axis side superimposed voltage Vhd) and a q-axis direction component Vhq (hereinafter referred to as a q-axis side superimposed voltage Vhq), and a d-axis A value obtained by adding (superimposing) the side superimposed voltage Vhd to the d-axis voltage FB command value Vfbd is calculated as the d-axis voltage command value Vd_c that is input to the dq-3 phase converter 26. A q-axis that calculates a value obtained by adding (superimposing) the q-axis side superimposed voltage Vhq to the q-axis voltage FB command value Vfbq as the q-axis voltage command value Vq_c that is input to the dq-3 phase converter 26. And a side superposition unit 29.

In the following description, the dq voltage vector (Vhd, Vhq) ^t expressing the periodic signal on the dq coordinate system may be referred to as a dq superimposed voltage vector Vhdq.

  In the present embodiment, the periodic signal generator 27, the d-axis side superimposing unit 28, and the q-axis side superimposing unit 29 realize the high-frequency superimposing means in the present invention.

  In the present embodiment, the periodic signal output from the periodic signal generation unit 27 has n (n ≧ 2) continuous control cycles as one period, and the dq superimposed voltage vector Vhdq (each The time series of the dq superimposed voltage vector Vhdq) for each control cycle is a periodic signal that satisfies the following equation (5). Therefore, in the periodic signal, the sum of the dq superimposed voltage vectors Vhdq of each control cycle in the period of one cycle becomes “0” (the average value of the voltages of the periodic signals becomes “0” in the one cycle). ) Is a periodic signal set as follows.

  Vhd (j) and Vhq (j) (j = 1, 2,..., N) are values of Vhd and Vhq in the jth control cycle within one period of the periodic signal.

  In the present embodiment, the periodic signal output by the periodic signal generator 27 is a periodic signal set in a pattern (time series pattern) as shown in FIG. In this example, the dq superimposed voltage vector Vhdq has three control cycles as one cycle (n = 3), and each control cycle (control cycle at times T11, T12, and T13) has a period signal. ) Is a periodic signal set in such a pattern that the sum of Vhd and the sum of Vhq becomes “0”. In addition, each arrow attached | subjected with the number of the circle of Fig.2 (a) has shown the dq superposition voltage vector Vhdq in each control cycle (control cycle of time T11, T12, T13) of one period of a periodic signal. The numbers with circles indicate the generation order of the superimposed voltage vector Vhdq in each control cycle of one period of the periodic signal.

  Supplementally, the pattern of the periodic signal is not limited to the pattern shown in FIG. For example, a periodic signal pattern may be set as shown in FIG. In this example, the periodic signal has four control cycles as one period, and the sum of the dq superimposed voltage vectors Vhdq for every two consecutive control cycles in the one period (Vhd in two control cycles at times T21 and T22, The sum of the respective values of Vhq and the sum of the respective values of Vhd and Vhq in the two control cycles at times T23 and T24) are set in a pattern that is “0”. In addition, each arrow which attached | subjected the circled number of FIG.2 (b), and the meaning of this circled number are the same as that of the case of Fig.2 (a).

  In this embodiment, a time series of the dq superimposed voltage vector Vhdq of the periodic signal having a pattern as shown in FIG. 2A is periodically output from the periodic signal generator 27. The d-axis voltage command value Vd_c is determined by adding the d-axis side superimposed voltage Vhd of the dq superimposed voltage vector Vhdq to the d-axis voltage FB command value Vfbd by the d-axis side superposition unit 28. The q-axis voltage command value Vq_c is determined by adding the q-axis side superimposed voltage Vhq of the dq superimposed voltage vector Vhdq of the periodic signal to the q-axis voltage FB command value Vfbq by the q-axis side superposition unit 29.

  Therefore, in order to converge the phase current of the armature 2 of the motor 1 to the target current, the dq as the dq voltage vector Vdq representing the drive voltage to be applied to the armature 2 of each phase in the dq coordinate system. The dq superimposed voltage vector Vhdq is superimposed on the voltage FB command vector Vfbdq. As a result, a set of the d-axis voltage command value Vd_c and the q-axis voltage command value Vq_c that defines the drive voltage actually applied to the armature 2 (command value of the dq voltage vector Vdq) is determined.

  Note that, when the dq superimposed voltage vector Vhdq is superimposed on the dq voltage FB command vector Vfbdq as described above, the arithmetic processing unit 11 superimposes the dq voltage FB command on which the dq superimposed voltage vector Vhdq is superimposed in each period of the periodic signal. The d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq of the vector Vfbdq are held at a constant value. In other words, the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq of the dq voltage FB command vector Vfbdq are updated for each cycle of the periodic signal (every three control cycles in this embodiment).

  Then, the arithmetic processing unit 11 uses the dq-3 phase conversion unit 26 to set the three-phase phase voltage command values Vu_c, Vv_c to the set of the q-axis voltage command value Vq_c and the d-axis voltage command value Vd_c determined as described above. , Vw_c and output to the PDU 13. At this time, the drive voltages of the phase voltage command values Vu_c, Vv_c, and Vw_c defined by the command value of the dq voltage vector Vdq obtained by adding the dq superimposed voltage vector Vhdq to the dq voltage FB command vector Vfbdq Applied to the child 2. In this case, the drive voltage applied to the armature 2 of each phase of the motor 1 is a drive voltage corresponding to the dq voltage FB command vector Vfbdq (a drive voltage obtained by converting Vfbdq by the dq-3 phase converter 26). As a center, a voltage that slightly fluctuates in the same cycle as the periodic signal is obtained.

  Supplementally, superimposing the dq superimposed voltage vector Vhdq representing the periodic signal in the dq coordinate system on the dq voltage FB command vector Vfbdq means that the dq voltage FB command vector Vfbdq is converted by the dq-3 phase converter 26. The phase voltage command value of each phase is converted to a periodic signal of each phase obtained by converting the dq superimposed voltage vector Vhdq by the dq-3 phase conversion unit 26 (the sum of voltage values in this cycle is also “0”). It is equivalent to superimposing “become”. Therefore, the superimposition of the periodic signal may be performed on the output side instead of on the input side of the dq-3 phase conversion unit 26.

  Further, the arithmetic processing unit 11 further superimposes the dq superimposed voltage vector Vhdq of the periodic signal on the dq voltage FB command vector Vfbdq as described above, with the three-phase-dq converting unit 21 and the dq-3 phase converting unit 26 A rotor angle estimator 30 is provided for calculating an estimated value θ ^ of the rotor angle used for the conversion process.

  Here, an outline of the rotor angle estimation processing by the rotor angle estimation unit 30 will be described below.

  A model that expresses the relationship between current and voltage in a dq coordinate system (hereinafter referred to as an estimated dq coordinate system) in which each axial direction is specified using the estimated value θ ^ of the rotor angle, the electrical angular velocity of the rotor is “0”. "Or a value close thereto, and when the voltage drop due to the resistance of the d-axis armature and the q-axis armature is at a sufficiently small level, it is expressed by the following equation (6).

  In this equation (6), Id ^ is the d-axis current in the estimated dq coordinate system, Iq ^ is the q-axis current in the estimated coordinate system, Vd ^ is the d-axis voltage in the estimated dq coordinate system, and Vq ^ is the estimated dq coordinate. The q-axis voltage, θe, in the system is the phase difference (= θ−θ ^) between the actual value θ of the rotor angle and the estimated value θ ^.

  Further, L0 and L1 in the equation (6) are expressed by the following equation (7) with respect to the average inductance Ld of the d-axis armature on the actual d-axis and the average inductance Lq of the q-axis armature on the actual q-axis. ) Is a variable defined as having a relationship.

The above equation (6) is expressed in a discrete time system, and the dq superimposed voltage vector Vhdq is superimposed on the dq voltage FB command vector Vfbdq as a dq voltage vector (Vd ^, Vq ^) ^t in each control cycle. The following equation (8) is obtained by using the following dq voltage vector (= Vfbdq + Vhdq).

  Note that Mdq in equation (8) is the same matrix as the matrix on the right side of equation (6) as shown in equation (9). Further, dId (i) and dIq (i) are respectively the first-order difference of the d-axis current between the i-th control cycle and the next control cycle, and the q-axis current as shown in the equation (10). It is a floor difference. Vfbdq (i) is a dq voltage FB command vector Vfbdq in the i-th control cycle, and Vhdq (i) is a dq superimposed voltage vector Vhdq in the i-th control cycle. Note that ΔT in Equation (10) is the time of one control cycle.

  Then, from the matrix defined by the equations (2) and (3) and the above equation (8) (more specifically, n equations (8) in each control cycle from i + 1 to i + n) (12) is obtained.

  Here, the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq of the dq voltage FB command vector Vfbdq in a period including n control cycles from the (i + 1) th to the (i + n) th control cycle are kept constant (Vfbdq (i + 1) = Vfbdq (i + 2) =... = Vfbdq (i + n)), and the dq superimposed voltage vector Vhdq of the periodic signal in this period is set to satisfy the above equation (5). The first term on the right side of the above equation (12) is “0”. Therefore, the following equation (13) is obtained.

  The left side of the formula (13) is the same as the right side of the formula (1). In this case, as apparent from comparison between the above equation (13) and the above equation (1), Vsdq on the left side of the above equation (1) is the phase difference as an error of the estimated value θ ^ of the rotor angle. It has a meaning as a value that changes (in proportion) according to a sinusoidal function value sin2θe of a double angle of θe. Further, Vcdq on the left side of the equation (1) has a meaning as a value that changes (in proportion) according to a cosine function value cos2θe of a double angle of the phase difference θe. Further, L0 on the left side of the equation (1) has a meaning as a value that changes according to the average inductances Ld and Lq of the d-axis armature and the q-axis armature as described above. Hereinafter, Vsdq may be referred to as a sine reference value and Vcdq may be referred to as a cosine reference value.

  Therefore, the rotor angle estimation unit 30 includes the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq of the dq voltage FB command vector Vfbdq in a period including n consecutive (three in this embodiment) control cycles. In a state where the dq superimposed voltage vector Vhdq of the periodic signal satisfying the above equation (5) is superimposed on the dq voltage FB command vector Vfbdq, and the control cycle in the period and the next The first-order difference between the d-axis current Id and the q-axis current Iq during the control cycle is observed. Then, the rotor angle estimation unit 30 performs the calculation of the left side of Expression (13), that is, the calculation of the right side of Expression (1) from the observed values of the first-order differences, thereby obtaining the expression (1). Vsdq, Vcdq, and L0 on the left side are calculated.

  In this embodiment, the rotor angle estimation unit 30 sets the value of L0 calculated by, for example, the calculation of the right side of the equation (1) as the inductance reference value in the present invention. Further, the rotor angle estimation unit 30 uses, for example, the sine reference value Vsdq calculated by the calculation of the right side of the equation (1) as the phase difference basic reference value in the present invention (hereinafter referred to as the reference symbol ΘEb). A value Vsdq / L0 obtained by dividing by the inductance reference value L0 is set.

  Here, the phase difference basic reference value ΘEb (= Vsdq / L0) calculated in this way is a state in which no magnetic saturation of the motor 1 occurs (for example, the torque command value Tr_c of the motor 1 is 0 [N · m]). 3), the waveform of the change in the value of ΘEb with respect to the phase difference θe is a waveform that is approximately proportional to the sine function value sin2θe of the double angle 2θe, as illustrated in the graph of FIG. However, in a state where the magnetic saturation of the motor 1 is occurring (for example, a state where the torque command value Tr_c of the motor 1 is set to the maximum torque), the waveform of ΘEb is magnetic saturation as illustrated in the graph of FIG. It becomes a waveform collapsed from a sinusoidal waveform (a waveform shown in the graph of FIG. 3A) in a state where no occurrence occurs. And this waveform collapse becomes so remarkable that the output torque of the motor 1 is large (as a result, the phase current which flows into the armature 2 is large). Furthermore, due to the occurrence of magnetic saturation, an offset of the level of the entire waveform of ΘEb also occurs. For example, the waveform of ΘEb illustrated in the graph of FIG. 3B is offset to the negative side with respect to the waveform of ΘEb illustrated in FIG.

Note that the graphs of FIGS. 3A and 3B are graphs of actual measurement data when the actual value of the rotor angle is set to a predetermined value. In addition, the tendency that the waveform of the phase difference basic reference value ΘEb collapses as described above in a state where magnetic saturation occurs is that Vsdq or Vsdq / √ (Vsdq / L0 exemplified above as Vsdq / L0 as ΘEb. The same applies when Vsdq ² + Vcdq ² ) is used.

  On the other hand, paying attention to the inductance reference value L0 calculated by the above equation (1), the inductance reference value L0 is in a state in which the magnetic saturation of the motor 1 occurs, for example, as shown in the graph of FIG. It has a tendency to change with respect to θe. As shown in the figure, L0 has linearity within a range where θe is near “0” even when magnetic saturation of the motor 1 occurs.

  Therefore, by linearly combining the phase difference basic reference value ΘEb and the inductance reference value L0 using an appropriate weighting factor, or by adding a more appropriate correction term to the linearly combined value, It is considered that a reference value can be generated such that the waveform of the change with respect to the phase difference θe is kept substantially constant regardless of the state of occurrence of magnetic saturation of the motor 1.

  Therefore, the rotor angle estimation unit 30 according to the present embodiment uses the inductance reference value L0 and the phase difference basic reference value ΘEb to calculate the double angle 2θe of the rotor angle θe by the calculation of the following equation (14). The phase difference estimated value ΘE whose value changes is determined.

ΘE = α · L0 + β · ΘEb + γ (14)

That is, the rotor angle estimator 30 uses α and β as weighting factors respectively applied to the inductance reference value L0 and the phase difference basic reference value ΘEb, and a value obtained by linearly combining these L0 and ΘEb (= α · L0 + β). A value obtained by adding a correction term γ to ΘEb) is calculated as a phase difference estimated value ΘE.

  In this case, as the phase current flowing through the armature 2 of the motor 1 increases and the magnetic saturation of the motor 1 is more intense, the above-described waveform collapse of ΘEb becomes more prominent. Therefore, the rotor angle estimation unit 30 sets the weight coefficients α and β and the value of the correction term γ so as to change according to the output torque of the motor 1 (the torque command value Tr_c in the present embodiment).

  Then, the rotor angle estimation unit 30 uses the phase difference estimated value ΘE obtained as described above, and the rotor angle estimated value θ ^ and the change speed of the rotor angle are estimated by the observer represented by the following equation (15). Calculate the estimated angular velocity ω ^.

  In this equation (15), Δt is the time of the update cycle of the estimated value θ ^ of the rotor angle, k is the update number of θ ^, and K1 and K2 are preset gain coefficients.

  The observer represented by the equation (15) converges the estimated value θ ^ of the rotor angle to Δt so that the estimated phase difference value ΘE converges to “0” and eventually the phase difference θe converges to “0”. It is calculated while sequentially updating in cycles.

  The above is the outline of the processing of the rotor angle estimation unit 30. In this embodiment, the rotor angle estimation unit 30 implements the rotor angle estimation means in the present invention. In this case, the rotor angle estimation unit 30 includes functions as basic reference value calculation means and phase difference estimation value calculation means in the present invention.

  More specifically, the processing of the arithmetic processing unit 11 including the processing of the rotor angle estimation unit 30 is executed as shown in the flowchart of FIG. The process of this flowchart is a process executed sequentially in the control cycle.

  In the following, in STEP 1, first, the arithmetic processing unit 11 takes in the U-phase current detection value Iu_s and the W-phase current detection value Iw_s, which are the outputs of the U-phase current sensor 12 u and the W-phase current sensor 12 w, respectively.

  Next, in STEP 2, the arithmetic processing unit 11 determines the first-order difference dIu of the U-phase current detection value Iu_s and the first-order difference of the W-phase current detection value Iw_s during one control cycle from the previous control cycle to the current control cycle. dIw is calculated by the rotor angle estimation unit 30. Specifically, the rotor angle estimator 30 subtracts the value obtained by subtracting the previous value (value acquired in the previous control cycle) from the current value (value acquired in the current control cycle) of the U-phase current detection value Iu_s. A value obtained by dividing by the control cycle time ΔT (that is, the amount of change in the U-phase current detection value Iu_s per control cycle time) is calculated as the U-phase side first-order difference dIu. The same applies to the first-order difference dIw on the W phase side. A value obtained by multiplying the change amount of the U-phase current detection value Iu_s and the change amount of the W-phase current detection value Iw_s per control cycle time by a predetermined gain (for example, ΔT), respectively, is used as the first-order difference dIu, It may be calculated as dIw.

Next, in STEP 3, the rotor angle estimation unit 30 of the arithmetic processing unit 11 converts the first-order difference dId of the d-axis current from the first-order differences dIu and dIw by a conversion process similar to that of the three-phase-dq conversion unit 21. , And a first-order difference dIq of the q-axis current are calculated. That is, the rotor angle estimation unit 30 multiplies a vector (dIu, dIw) ^t having components dIu and dIw by a transformation matrix, thereby obtaining a vector (dId, dIq) ^t (hereinafter referred to as a first-order differential current vector dIdq). Is calculated. In this case, the component value of each row and each column of the transformation matrix is determined according to the latest value of the estimated value θ ^ of the rotor angle for dq transformation updated in STEP 9 described later.

  Next, in STEP 4, the arithmetic processing unit 11 determines whether or not the current value of the counter variable ptr is “1”. At this time, when ptr = 1, the arithmetic processing unit 11 executes the processing of STEPs 5 to 9, and when not ptr = 1, the arithmetic processing unit 11 executes the processing from STEP 10.

  Here, the counter variable ptr is a variable that is incremented for each control cycle in STEP 15 described later, and is reset to “0” when the incremented ptr value is “3”. For this reason, every time three control cycles corresponding to one period of the periodic signal elapse, the determination result of STEP 4 becomes affirmative and the processes of STEP 5 to 9 are executed. In the following description, the control cycles in which the value of the counter variable ptr at the start of each control cycle is “0”, “1”, “2” are the control cycle (0), the control cycle (1), and the control cycle, respectively. Sometimes referred to as cycle (2).

  When the determination result of STEP 4 is ptr = 1 (when the current control cycle is the control cycle (1)), the arithmetic processing unit 11 performs the processing of STEP 5 to 7 by the rotor angle estimation unit 30. . In this case, first, in STEP 5, the rotor angle estimation unit 30 calculates the sine reference value by performing the calculation of the right side of the equation (1) (more specifically, the equation in which n in the equation (1) is n = 3). Vsdq, cosine reference value Vcdq, and inductance reference value L0 are calculated.

  Specifically, the rotor angle estimation unit 30 calculates the first-order differential current vector dIdq calculated in the previous control cycle (2), the first-order differential current vector dIdq calculated in the previous control cycle (0), The first-order differential current vector dIdq calculated in the control cycle (1) is converted into dIdq (i + 1), dIdq (i + 2), dIdq (i + 3) (= dIdq (i + n) in equation (1), respectively. )), The sine reference value Vsdq, the cosine reference value Vcdq, and the inductance reference value L0 are calculated by executing the calculation of the left side of the equation (1). In this case, dq superimposed voltage vectors Vhdq (i + 1), Vhdq (i + 2), Vhdq (i + 3) (= Vhdq (i + n) that define each component value of the matrix Cdq in the equation (1). )) Are the values of the dq superimposed voltage vector Vhdq output by the periodic signal generator 27 in the control cycle (1), control cycle (2), and control cycle (0), respectively. In this embodiment, as will be described later, the voltage component corresponding to the dq superimposed voltage vector Vhdq output by the periodic signal generator 27 in each control cycle is applied to the armature 2 of each phase of the next control cycle motor 1. Is done.

  Next, in STEP 6, the rotor angle estimating unit 30 calculates the phase difference basic reference value ΘEb as described above from the sine reference value Vsdq calculated as described above and the inductance reference value L0. In other words, a value obtained by dividing Vsdq by L0 (= Vsdq / L0) is calculated as ΘEb. In this embodiment, since the cosine reference value Vcdq is not used, the calculation process of the cosine reference value Vcdq may be omitted in STEP5.

  Further, in STEP7, the rotor angle estimation unit 30 calculates the phase difference estimated value ΘE by the equation (14) using the inductance reference value L0 calculated as described above and the phase difference basic reference value ΘEb.

  In this case, the rotor angle estimator 30 calculates the values of the weighting coefficients α and β and the correction term γ used in the calculation of Expression (14) from the torque command value Tr_c, as shown in FIGS. The determination is made based on a preset data table as shown in FIG. FIGS. 6A, 6B, and 6C are data tables corresponding to the weighting factor α, the weighting factor β, and the correction term γ, respectively.

  The values of α, β, and γ in these data tables are the waveforms of the phase difference estimated value ΘE calculated by the calculation of the equation (14) (with respect to the phase difference θe) at each torque command value Tr_c at the square points in the figure. (The waveform of change in ΘE) is identified by a method such as a least square method so as to match the target waveform. In this case, the target waveform is, for example, a waveform of the phase difference basic reference value ΘEb in a state where the torque command value Tr_c is 0 [N · m] (a state where no magnetic saturation of the motor 1 occurs). . Accordingly, the values of α, β, and γ when Tr_c = 0 [N · m] are set to “0”, “1”, and “0”, respectively. Therefore, when Tr_c = 0 [N · m], the phase difference estimated value ΘE matches the phase difference basic reference value ΘEb.

  Note that the values of α, β, and γ corresponding to the torque command values Tr_c other than the torque command values Tr_c at the square points in the figure are determined by linear interpolation, for example.

  The rotor angle estimator 30 calculates the phase difference estimated value ΘE through the processing in STEPs 5 to 7 described above.

Next, proceeding to STEP 8, the arithmetic processing unit 11 converts the set of phase current detection values Iu_s and Iw_s taken in STEP 1 by the three-phase-dq conversion unit 21 as described above, so that the d-axis for current feedback control is obtained. A set of the detected current value Id_s and the detected q-axis current value Iq_s (detected value (Id_s, Iq_s) ^{t of the} dq current vector Idq) is calculated and held. The detected value of the current vector Idq calculated in STEP 8 is processed in STEP 11 (to be described later) in the control cycle (0) of the next time (next next time) (the d-axis current deviation calculation unit 23, q-axis current deviation calculation). Used by the unit 24 and the current FB control unit 25).

  Note that the estimated value θ ^ of the rotor angle used in the process of the three-phase-dq conversion unit 21 in STEP 8 is the same as the value of θ ^ used in the process of STEP 3 of the current control cycle (1).

  Next, in STEP 9, the arithmetic processing unit 11 determines the value of the rotor angle estimation value θ ^ for dq conversion (specifically, the processing of the 3-phase-dq conversion unit 21 and the dq-3 phase conversion unit 26 and the processing of STEP 3) Is updated to a value newly determined by the rotor angle estimation unit 30 in STEP 12 described later in the previous control cycle (0). Therefore, the value of θ ^ used in the processing of STEP 3 is kept constant during the period of three control cycles from the control cycle (2) to the control cycle (1). Further, the value of θ ^ used for the processing of the three-phase-dq conversion unit 21 of STEP 8 in the control cycle (1) in the period of the three control cycles is used in the processing of STEP 3 in the period of the three control cycles. It becomes the same as the value of θ ^.

  After performing the processing of STEP 9 as described above, or when ptr ≠ 1 in STEP 4, the arithmetic processing unit 11 determines whether or not the current value of the counter variable ptr is “0” in STEP 10. to decide.

  At this time, when ptr = 0, the arithmetic processing unit 11 sequentially executes the processing of STEPs 11 and 12, and when ptr ≠ 0, the arithmetic processing unit 11 executes the processing from STEP13.

  In the case where ptr = 0 in STEP 10 (when the current control cycle is the control cycle (0)), first, in STEP 11, the arithmetic processing unit 11 performs the d-axis current deviation calculation unit 23, the q-axis current deviation in step 11. By executing the processing of the calculation unit 24 and the current FB control unit 25, a set of the d-axis voltage FB command value Vfbd and the q-axis voltage FB command value Vfbq (dq voltage FB command vector Vfbdq) is calculated.

In this case, the d-axis current deviation ΔId and the q-axis current deviation ΔIq necessary for calculating the dq voltage FB command vector Vfbdq are detected as the current vector Idq calculated and held in STEP 8 of the previous control cycle (1). value (Id_s, Iq_s) is calculated from the ^t, the command value of the dq current vector Idq (Id_c, Iq_c) second preceding value of ^t and.

  Next, proceeding to STEP 12, the arithmetic processing unit 11 causes the rotor angle estimation unit 30 to update the estimated value θ ^ of the rotor angle according to the above equation (15). In this case, the value calculated in STEP 7 of the previous control cycle (1) is used as the phase difference estimated value θE necessary for updating the estimated value θ ^ of the rotor angle.

As described above, in the control cycle (0) in which ptr = 0, the dq voltage FB command vector Vfbdq (= (Vfbd, Vfbq) ^t ) is updated and the estimated value θ ^ of the rotor angle is updated. . Therefore, the dq voltage FB command vector Vfbdq and the estimated value θ ^ of the rotor angle are updated every three control cycles. Then, the dq voltage FB command vector Vfbdq and the estimated value θ ^ of the rotor angle are constant in a period until the next control cycle (0) after three control cycles after being updated in STEP12 of the control cycle (0). Kept at the value.

  Since the process of STEP 9 is included, the estimated rotor angle θ ^ updated in STEP 12 is used in the processes of STEPs 3 and 8 until the process of STEP 9 of the next control cycle (1) is executed. First, in the period of three control cycles from the next control cycle (2) to the control cycle (1), it is used in the processing of STEP 3 and in the control cycle (1) of the period of the three control cycles. , Used in the processing of STEP 8 (processing of the three-phase-dq conversion unit 21).

After executing the processing of STEP12 as described above or when ptr ≠ 0 in STEP10, the arithmetic processing unit 11 in STEP13 determines that the periodic signal generating unit 27, the d-axis side superimposing unit 28, and q The axis-side superimposing unit 29 superimposes (adds) the dq superimposed voltage vector Vhdq (= (Vhd, Vhq) ^t ) of the periodic signal to the current value of the dq voltage FB command vector Vfbdq (= (Vfbd, Vfbq) ^t ). Thus, a set of the d-axis voltage command value Vd_c and the q-axis voltage command value Vq_c (command value (Vd_c, Vq_c) ^t ) of the dq voltage vector Vdq) is calculated.

In this case, dq superposed voltage vector (Vhd, ^Vhq) dq voltage FB command vector for superimposing the ^t (Vfbd, ^Vfbq) the current value of ^t, calculated in STEP11 latest control cycle (0) to be ptr = 0 Value. In addition, the periodic signal generator 27 generates a dq superimposed voltage vector Vhdq of the periodic signal with a period of three control cycles from the control cycle (0) to the control cycle (2) as one period. Since the sum of the dq superimposed voltage vector Vhdq in this period of one cycle is 0, the average value of the command values (Vd_c, Vq_c) ^t of the dq voltage vector Vdq in this period is a constant value in that period. It matches the voltage FB command vector Vfbdq.

Next, proceeding to STEP 14, the arithmetic processing unit 11 uses the dq-3 phase conversion unit 26 to set the command value (Vd_c, Vq_c) ^t of the dq voltage vector Vdq obtained in STEP 13 to the set of phase voltage command values Vu_c, Vv_c, Vw_c. Convert to

  Note that the estimated value θ ^ of the rotor angle used in the processing of the dq-3 phase conversion unit 26 in STEP 14 was updated in STEP 9 in the latest (most recent) control cycle (1) where ptr = 1. Value. .

  The phase voltage command values Vu_c, Vv_c, and Vw_c calculated in STEP 14 are output to the PDU 13 in the next control cycle. Therefore, the drive voltage of the phase voltage command values Vu_c, Vv_c, Vw_c calculated in STEP 14 is actually applied to the armature 2 of each phase of the motor 1 by the PDU 13 with a delay of one control cycle.

  Next, proceeding to STEP 15, the arithmetic processing unit 11 increments the value of the counter variable ptr by 1. Further, the arithmetic processing unit 11 determines whether or not the incremented ptr value is “3” in STEP 16. Then, when ptr = 3, the arithmetic processing unit 11 resets the value of ptr to “0” in STEP 17 and ends the process of the current control cycle. If ptr ≠ 3, the process of the current control cycle is terminated as it is.

  The above is the details of the processing of the arithmetic processing unit 11 of the control device 10.

  In the present embodiment, the basic reference value calculation means in the present invention is realized by the processing of STEPs 2, 3, 5, and 6 performed by the rotor angle estimation unit 30. Further, the processing of STEP 7 realizes the phase difference estimated value calculation means in the present invention.

By the processing described above, the component corresponding to the dq voltage FB command value (Vfbd, Vfbq) ^t included in the drive voltage actually applied to the armature 2 of each phase of the motor 1 is controlled from the control cycle (1). It is maintained constant in a period of three control cycles up to cycle (0) (period of one period of the periodic signal). In this period of three control cycles, the voltage component corresponding to the periodic signal (dq superimposed voltage vector (Vhd, Vhq) ^t ) included in the drive voltage actually applied to the armature 2 of each phase of the motor 1. ) Changes for each control cycle within the period, with the period as one period. However, the average value of the drive voltage actually applied to the armature 2 of each phase during the period is a constant value corresponding to the dq voltage FB command vector (Vfbd, Vfbq) ^t . Therefore, the average value of the phase current of the armature 2 of each phase of the motor 1 during the period is controlled so as to match the current value corresponding to the command value (Id_c, Iq_c) ^t of the dq current vector. Become.

  As described above, the d-axis current detection value Id_s and the q-axis current detection value Iq_s are obtained in a state where the drive signal including the high-frequency periodic signal is superimposed on the drive voltage applied to the armature 2 of each phase of the motor 1. First-order differences dId, dIq (current change between each control cycle from control cycle (1) to control cycle (0) and the next control cycle) are observed. From these observation values, the phase difference basic reference value ΘEb and the inductance reference value L0 are calculated by the rotor angle estimation unit 30 as described above. Further, using this ΘEb and L0, the phase difference estimated value ΘE is calculated by the equation (14). Further, the estimated value θ ^ of the rotor angle is updated from the estimated phase difference value ΘE by the observer calculation shown in the equation (15).

  As a result, the estimated value θ ^ of the rotor angle is updated every three control cycles so that the estimated phase difference value ΘE converges to “0” and consequently the phase difference θe converges to “0”.

  Here, in the present embodiment, the phase difference estimated value ΘE used for the operation of the observer, in other words, the phase difference estimated value ΘE that defines the update amount of the rotor angle estimated value θ ^ is expressed by the equation (14). Calculated by calculation. In this case, the weighting factors α and β and the correction term γ are determined according to the torque command value Tr_c using the data table set as described above.

  Therefore, in the calculated phase difference estimated value ΘE, the magnetic saturation of the motor 1 occurs regardless of whether the waveform of the change in the value of ΘE with respect to the phase difference θe is the actual magnetic saturation occurrence state of the motor 1. The value is such that the waveform is maintained in a non-existing state (a sinusoidal waveform substantially proportional to sin2θe). That is, the value of ΘE coincides with or substantially coincides with a value obtained by multiplying sin2θe by a predetermined proportionality constant.

  Actually, according to the simulation result by the present inventor, it has been confirmed that the waveform of ΘE becomes, for example, a waveform as illustrated in the graphs of FIGS. 7 (a), (b), and (c). Here, FIGS. 7A to 7C show a state in which the actual value of the rotor angle is set to a predetermined value and the torque command value Tr_c is set to a different value (as a result, the state of occurrence of magnetic saturation of the motor 1). Is a graph showing the waveform of ΘE in a state where the values are different from each other. In this case, the waveform of ΘE shown in FIG. 7A is a waveform in a state where the torque command value Tr_c is 0 [N · m] (a state where the magnetic saturation of the motor 1 has not occurred). 7B and 7C show the state where the torque command value Tr_c is half the maximum torque and the maximum torque respectively (these states are states in which the magnetic saturation of the motor 1 occurs. It is a waveform of ΘE in (A).

  As shown in these graphs, the waveform of the change in the phase difference estimated value ΘE with respect to the phase difference θe is substantially constant regardless of the torque command value Tr_c and, consequently, regardless of the state of occurrence of magnetic saturation of the motor 1. It can be seen that the sine waveform is maintained. In this case, since the phase difference estimated value ΘE is substantially proportional to sin2θe, when the phase difference θe is “0”, ΘE≈0. In addition, the value of ΘE is opposite in polarity and symmetrical between the case where the phase difference θe is positive and the case where it is negative.

  Although illustration is omitted, in the state where the torque command value Tr_c is set to each of the above values (0 [N · m], half of the maximum torque, maximum torque), the actual value of the rotor angle is changed to various other values. Even when the angle is set, it has been confirmed that the waveform of ΘE is substantially the same as each of the waveforms shown in FIGS. In addition to the above values (0 [N · m], half of the maximum torque, maximum torque), the torque command value Tr_c is, for example, 25%, 75%, 10%, 30%, 60% of the maximum torque, Even when the value was set to 80%, it was confirmed that the waveform of ΘE was maintained in a sinusoidal waveform similar to that shown in FIGS. In this case, the values of α, β, and γ when Tr_c is set to 10%, 30%, 60%, and 80% of the maximum torque are obtained from the data tables shown in FIGS. 6 (a) to 6 (c). It is determined by linear interpolation.

  As described above, the waveform of the change in the estimated phase difference value ΘE with respect to the phase difference θe is maintained in a substantially constant sinusoidal waveform regardless of the state of occurrence of magnetic saturation in the motor 1. For this reason, the estimated value θ ^ of the rotor angle updated / calculated by the observer of equation (15) can be accurately and stably converged to the actual value regardless of the state of occurrence of magnetic saturation of the motor 1. It will be possible. That is, the rotor angle can be estimated accurately and stably regardless of the state of occurrence of magnetic saturation of the motor 1.

  Further, since the rotor angle can be estimated accurately and stably, the reluctance torque resulting from the saliency of the rotor can be used effectively, and the desired torque of the motor 1 can be generated efficiently.

  The simulation result verified about the above effects in this embodiment is shown to Fig.8 (a), (b). In addition, for comparison with this, the simulation results when the conventional rotor angle estimation method is used are shown in FIGS. 9 (a) and 9 (b).

  FIG. 8A is a graph showing a change in the estimated value θ ^ of the rotor angle calculated according to the present embodiment and a change in the actual value θ of the rotor angle by a solid line and a broken line, respectively (in the figure, a solid line) And the broken line almost overlap). The energization current of the armature 2 of the motor 1 is a current set so that the output torque of the motor 1 is almost the maximum torque. The change speed of the actual value θ of the rotor angle, that is, the electrical angular speed of the rotor is a constant value.

  As shown in the drawing, the estimated value θ ^ of the rotor angle changes following the actual value θ while matching the actual value θ with high accuracy. Thus, in this embodiment, the estimated value θ ^ of the rotor angle can be matched with the actual value θ with high accuracy.

  FIG. 8B is a graph showing a change in the actual output torque of the motor 1. As shown in the figure, the output torque is kept substantially constant at a torque near the maximum torque.

  On the other hand, FIG. 9A shows a temporal change in the estimated value θ ^ of the rotor angle calculated using a conventional rotor angle estimation method and a temporal change in the actual value θ of the rotor angle. Are graphs indicated by a solid line and a broken line, respectively. Here, the conventional rotor angle estimation method uses a predetermined offset value (−0.25) for the phase difference basic reference value ΘEb (= Vsdq / L0) instead of the phase difference estimation value ΘE in the present embodiment. The rotor angle is estimated using the added value. The energization current of the motor 1 is set to be the same as in the case of FIG. Further, the change speed (electrical angular speed) of the actual value θ of the rotor angle is the same as that in FIG.

  In this case, as shown in FIG. 9A, the estimated value θ ^ of the rotor angle by the conventional rotor angle estimation method has an error larger than that shown in FIG. 8A with respect to the actual value θ. .

  In this case, the actual output torque of the motor 1 changes as shown in the graph of FIG. As can be seen by comparing the graph of FIG. 9B with the graph of FIG. 8B, the rotor angle is estimated by the method of this embodiment rather than the case of estimating the rotor angle by the conventional method. It can be seen that the average value of the output torque can be increased.

  Next, a second embodiment of the present invention will be described with reference to FIGS.

  In the present embodiment, the value used as the phase difference basic reference value ΘEb is different from that of the first embodiment. Accordingly, the same components as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof is omitted.

  In the first embodiment, the value Vsdq / L0 obtained by dividing the sine reference value Vsdq by the inductance reference value L0 is calculated as the phase difference basic reference value ΘEb. On the other hand, in this embodiment, the sine reference value Vsdq calculated by the equation (1) is used as it is as the phase difference basic reference value ΘEb.

  In this case, the waveform of the change in the phase difference basic reference value ΘEb (= Vsdq) with respect to the phase difference θe is roughly sinusoidal, but the waveform of the change in ΘEb (= Vsdq / L0) in the first embodiment. Is different. For example, the waveform of the change in the phase difference basic reference value ΘEb (= Vsdq) in the present embodiment has a larger amplitude or the like than the waveform of the change in ΘEb (= Vsdq / L0) in the first embodiment. There is a tendency to vary greatly depending on the state of occurrence of magnetic saturation.

  Therefore, in this embodiment, when the phase difference estimated value ΘE is obtained from the phase difference basic reference value ΘEb and the inductance reference value L0, the weighting coefficients α, β and the correction term γ in the linear combination are obtained from the motor 1. A data table different from that of the first embodiment is used as a data table for determining according to the output torque (torque command value Tr_c).

  Specifically, in this embodiment, the data tables for determining the weighting factors α and β and the correction term γ are set as illustrated in FIGS. 10A, 10B, and 10C, respectively. Has been.

  The values of α, β, and γ in these data tables are the phase differences calculated by the calculation of equation (14) at each torque command value Tr_c at the square points in the figure, as in the first embodiment. The waveform of the estimated value ΘE (the waveform of the change in ΘE with respect to the phase difference θe) is identified by a method such as the least square method so as to match the target waveform. In this case, in this embodiment, the target waveform is, for example, the waveform of the phase difference basic reference value ΘEb (= ΘE) in the state where the torque command value Tr_c is 0 [N · m] in the first embodiment. is there.

  As in the case of the first embodiment, α, β, γ corresponding to torque command values Tr_c other than the torque command values Tr_c at the respective square points in FIGS. 10A, 10B, and 10C. Is determined by linear interpolation, for example.

  The present embodiment is the same as the first embodiment except for the matters described above.

  This embodiment also has the same effect as the first embodiment.

  That is, the phase difference estimated value ΘE calculated in the present embodiment is such that the waveform of the change in the value of ΘE with respect to the phase difference θe does not depend on the actual magnetic saturation occurrence state of the motor 1 as in the first embodiment. , Sin2θe is a value that is maintained in a sinusoidal waveform that is substantially proportional.

  Actually, according to the simulation result by the present inventor, the waveform of the phase difference estimated value ΘE in the present embodiment is, for example, the waveform illustrated in the graphs of FIGS. 11 (a), (b), and (c). It was confirmed. Here, FIGS. 11 (a) to 11 (c) show the torque command value while setting the actual value of the rotor angle to a certain predetermined value, as in FIGS. 7 (a) to (c) according to the first embodiment. It is a graph which shows the waveform of (theta) E in the state which set Tr_c to a mutually different value. In this case, the waveform of ΘE shown in FIG. 11A is a waveform in a state where the torque command value Tr_c is 0 [N · m] (a state where the magnetic saturation of the motor 1 has not occurred). Further, the waveforms of ΘE shown in FIGS. 11B and 11C are states in which the torque command value Tr_c is set to half the maximum torque and the maximum torque, respectively (these states are states in which the magnetic saturation of the motor 1 occurs. It is a waveform of ΘE in (A).

  As shown in these graphs, as in the case of the first embodiment, the phase difference estimation with respect to the phase difference θe is performed regardless of the torque command value Tr_c, and thus regardless of the state of occurrence of magnetic saturation of the motor 1. It can be seen that the waveform of the change in the value ΘE is maintained in a substantially constant sinusoidal waveform. In this case, similarly to the first embodiment, the estimated phase difference value ΘE is substantially proportional to sin2θe, and therefore ΘE≈0 when the phase difference θe is “0”. In addition, the value of ΘE is opposite in polarity and symmetrical between the case where the phase difference θe is positive and the case where it is negative.

  Although illustration is omitted, in the state where the torque command value Tr_c is set to each of the above values (0 [N · m], half of the maximum torque, maximum torque), the actual value of the rotor angle is changed to various other values. Even when the angle is set, it has been confirmed that the waveform of ΘE is substantially the same as each of the waveforms shown in FIGS. In addition to the above values (0 [N · m], half of the maximum torque, maximum torque), the torque command value Tr_c is, for example, 25%, 75%, 10%, 30%, 60% of the maximum torque, Even when the value was set to 80%, it was confirmed that the waveform of ΘE was maintained in the same sinusoidal waveform as in FIGS. In this case, the values of α, β, and γ when Tr_c is set to 10%, 30%, 60%, and 80% of the maximum torque are obtained from the data table shown in FIGS. 10 (a) to 10 (c). It is determined by linear interpolation.

  As described above, also in the present embodiment, as in the first embodiment, the waveform of the change in the phase difference estimated value ΘE with respect to the phase difference θe is substantially constant regardless of the state of occurrence of magnetic saturation in the motor 1. A sinusoidal waveform is maintained. For this reason, the estimated value θ ^ of the rotor angle updated / calculated by the observer of equation (15) can be accurately and stably converged to the actual value regardless of the state of occurrence of magnetic saturation of the motor 1. It will be possible. That is, the rotor angle can be estimated accurately and stably regardless of the state of occurrence of magnetic saturation of the motor 1.

  Further, since the rotor angle can be estimated accurately and stably, the reluctance torque resulting from the saliency of the rotor can be used effectively, and the desired torque of the motor 1 can be generated efficiently.

In the first and second embodiments described above, the case where Vsdq / L0 or Vsdq is used as the phase difference basic reference value ΘEb has been described as an example, but other than those described above (for example, Vsdq / √ (Vsdq ² + Vcdq ² )) may be used as the phase difference basic reference value ΘEb.

  In each of the embodiments, the weighting factors α and β and the correction term γ are determined according to the torque command value Tr_c. However, according to other parameters that are substantially proportional to the output torque of the motor 1. Thus, the values of α, β, and γ may be variably determined. For example, the axial component (d-axis current in each of the embodiments) perpendicular to the magnetic field direction of the field pole of the rotor in the dq current vector when the current of the armature 2 of the motor 1 is expressed in the dq coordinate system is It is approximately proportional to the output torque of the motor 1. Therefore, the values of α, β, and γ may be determined according to the detected value or command value (target value) of the axial component orthogonal to the magnetic field direction of the rotor field pole in the dq current vector.

  In each of the above embodiments, the waveform of the phase difference basic reference value ΘEb (= Vsdq / L0) in the first embodiment in a state where the torque command value Tr_c is 0 [N · m] is used as the phase difference estimated value ΘE. However, the values of α, β, and γ may be determined using another waveform as the target waveform of ΘE. For example, a sine function value sin2θe which is a double angle of the phase difference θe, or a waveform obtained by multiplying this by a proportional constant (≠ 0) of an arbitrary predetermined value is used as a target waveform for ΘE, α, β, γ The value may be determined.

  Further, in the vector control in each of the above embodiments, the magnetic field direction of the field pole of the rotor is the q-axis direction, but the magnetic flux direction may be defined as the d-axis direction and the vector control calculation process may be executed. .

  DESCRIPTION OF SYMBOLS 1 ... DC brushless motor, 2u, 2v, 2w ... Armature, 10 ... Control device, 12u, 12w ... Current sensor (current detection means), 21 ... Three-phase-dq conversion part (dq basic voltage determination means), 23 ... d-axis current deviation calculating means (dq basic voltage determining means), 24... q-axis current deviation calculating means (dq basic voltage determining means), 25... current FB control section (dq basic voltage determining means), 27. (High frequency superimposing means), 28... D axis side superimposing section (high frequency superimposing means), 29... Q axis side superimposing section (high frequency superimposing means), 30... Rotor angle estimating section (rotor angle estimating means), STEPs 2, 3, 5 , 6 ... basic reference value calculation means, STEP 7 ... phase difference estimated value calculation means.

Claims (3)

Current detection means for detecting a phase current flowing in the armature of the DC brushless motor, rotor angle estimation means for estimating the rotor angle of the motor, detected value of phase current by the current detection means, and rotor by the rotor angle estimation means A d-axis voltage that defines a drive voltage to be applied to the armature in order to converge the detected value of the phase current to a target current by executing a predetermined feedback calculation process using the estimated value of the angle And dq basic voltage determining means for determining a basic value of q-axis voltage and a basic value of q-axis voltage, and at least a drive voltage defined by the basic value of the d-axis voltage and the basic value of the q-axis voltage is applied to the armature In the DC brushless motor control apparatus for controlling energization of the armature,
High-frequency superimposing means for applying to the armature a high-frequency superimposed drive voltage obtained by superimposing a high-frequency periodic signal on a drive voltage defined by the basic value of the d-axis voltage and the basic value of the q-axis voltage;
In the application state of the high-frequency superimposed drive voltage, changes in the d-axis current and the q-axis current corresponding to the change in the phase current of the armature are observed based on the detected value of the phase current and the estimated value of the rotor angle, Based on the observed values of the change in the d-axis current and the q-axis current, the value changes according to the double angle of the phase difference between the estimated value of the rotor angle and the actual value of the rotor angle, and A basic reference value calculating means for calculating a phase difference basic reference value whose waveform of change changes depending on the state of occurrence of magnetic saturation of the motor, and calculating an inductance reference value as a value depending on the average inductance of the motor;
Based on the phase difference basic reference value, at least the phase difference basic reference value and the inductance reference value are linearly combined to change the waveform of the change with respect to the phase difference depending on the state of magnetic saturation of the motor. A phase difference estimated value calculating means for calculating a phase difference estimated value that is also suppressed,
The said rotor angle estimation means updates the estimated value of the said rotor angle according to the said phase difference estimated value, The control apparatus of the DC brushless motor characterized by the above-mentioned. In the control apparatus of the DC brushless motor according to claim 1,
The phase difference estimated value calculating means outputs the output of the motor as at least one of a weighting coefficient applied to the phase difference basic reference value in the linear combination and a weighting coefficient applied to the inductance reference value. A control apparatus for a DC brushless motor, wherein a weighting factor set so that the value changes depending on torque is used. In the control apparatus of the DC brushless motor according to claim 1 or 2,
The phase difference estimated value calculation means adds a correction term set so that the value varies depending on the output torque of the motor to a value obtained by linearly combining the phase difference basic reference value and the inductance reference value. By doing so, the phase difference estimated value is calculated, and a control device for the DC brushless motor.

Images