JP2023169905A - Temperature estimation device of pm motor - Google Patents

Abstract

To provide a temperature estimation device of a PM motor capable of easily estimating a resistor temperature and a magnet temperature of the PM motor without adding a torque sensor, a voltage sensor or the like.SOLUTION: A temperature estimation device of a PM motor includes a parameter identification part 60 which calculates a current of a motor model using a designed motor model 61 with a motor detection current detecting a current flowing through a PM motor 1 driven by an inverter 2, an angular frequency obtained by detecting the rotation number of the motor 1, and a voltage command obtained by performing current control of a current controller 34 so that the motor detection current is equal to a current command as inputs and identifies motor parameters including a coil resistance and magnet magnetic flux of the motor by adjusting a coil resistance, magnet magnetic flux and an inductance of the motor model 61 using a parameter adjuster 63 so that difference between the motor model current and the motor detection current is eliminated ; and a temperature computation part 68 which estimates a resistor temperature and a magnet temperature from the identified coil resistance and magnet magnetic flux.SELECTED DRAWING: Figure 1

Description

The present invention relates to estimating the magnet temperature of a synchronous motor, and more particularly to a device for estimating the magnet temperature of a PM motor and the resistance temperature of a stator winding.

Conventionally, techniques for estimating the magnet temperature and stator winding resistance temperature of a motor have been proposed, for example, in Non-Patent Documents 1 and 2 and Patent Documents 1 to 4.

In Non-Patent Document 1, in order to improve the performance of speed sensorless control, a function of the square of the error of the stator winding resistance and magnet magnetic flux is used as an evaluation function (Lyapunov function), and the time derivative of the evaluation function is set to be negative. By changing the estimated values of the resistance and magnet magnetic flux, the magnet temperature is estimated based on the induced voltage measured by the voltage sensor.

In Non-Patent Document 2, the induction motor resistance is estimated by using the Lyapunov function.

Patent Document 1 proposes magnet temperature estimation using a torque sensor, and states that the advantage of using a torque sensor is that temperature can be estimated when the motor is not rotating. The details of the estimation method are described in paragraph numbers "0086" to "0092" of "Detailed explanation", and the current value and torque measurement value are substituted into the relational expression of current, magnetic flux, and torque to obtain the magnetic flux, and the change in magnetic flux is This method estimates the magnet temperature from

“Improving the performance of IPMSM position sensorless vector control by online simultaneous identification of stator resistance and permanent magnet flux linkage”, IEEJ Transactions D (Industrial Applications Journal) 129.7 (2009): 698-704. “Rotor resistance and stator resistance estimation method for induction motor control”, H24 National Conference of the Institute of Electrical Engineers of Japan 4-132.

Japanese Patent Application Publication No. 2019-170004 JP 2021-118652 Publication JP2021-016226A JP2021-090340A

Since the method using induced voltage as in Patent Document 2 needs to measure voltage, it is necessary to add a voltage sensor.

In a method based on a torque sensor such as that disclosed in Patent Document 1, it is necessary to add a torque sensor.

Since the method using an observer in Patent Document 3 requires a model, accurate parameters are required.

Although Patent Document 4 proposes state estimation using machine learning in a general form, it is not specific to magnet temperature, and does not clearly describe technology related to magnet temperature estimation.

In addition to the above, Patent Document 1, Patent Document 2, and Patent Document 4 do not estimate the temperature of the stator winding resistance. In Non-Patent Document 1, stator winding resistance and magnet magnetic flux are estimated, so it is possible to estimate temperature changes from these changes. This may cause errors. Non-Patent Document 2 targets induction machines.

The present invention solves the above problems, and its purpose is to provide a PM motor temperature estimation device that can easily estimate the resistance temperature and magnet temperature of the PM motor without adding a torque sensor, voltage sensor, etc. Our goal is to provide the following.

A PM motor temperature estimation device according to claim 1 for solving the above problem,
Current control is performed so that the motor detection current obtained by detecting the current flowing through the PM motor driven by the inverter, the angular frequency obtained by detecting the rotation speed of the motor, and the motor detection current become equal to the current command. Using the obtained voltage command as input, calculate the current of the motor model using the motor model, and adjust the winding resistance and magnet magnetic flux of the motor model so as to eliminate the difference between the calculated motor model current and the motor detection current. , a parameter identification unit that identifies motor parameters including motor winding resistance, magnet magnetic flux, and inductance by adjusting inductance;
The present invention is characterized by comprising a temperature calculation section that estimates a resistance temperature and a magnet temperature from the winding resistance and magnet magnetic flux identified by the parameter identification section.

The temperature estimating device for a PM motor according to claim 2 comprises:
Current control is performed so that the motor detection current obtained by detecting the current flowing through the PM motor driven by the inverter, the angular frequency obtained by detecting the rotation speed of the motor, and the motor detection current become equal to the current command. Using the obtained voltage command as input, calculate the current of the motor model using the motor model, and adjust the winding resistance and magnet magnetic flux of the motor model so as to eliminate the difference between the calculated motor model current and the motor detection current. a parameter identification unit that identifies motor parameters including motor winding resistance and magnet magnetic flux by adjusting the
The present invention is characterized by comprising a temperature calculation section that estimates a resistance temperature and a magnet temperature from the winding resistance and magnet magnetic flux identified by the parameter identification section.

The temperature estimating device for a PM motor according to claim 3 has the following features in claim 1 or 2:
The inverter drives the PM motor by vector control, and the vector control uses voltage and current in a dq coordinate system synchronized with the rotor of the PM motor.

The temperature estimating device for a PM motor according to claim 4 has the following features in claim 1:
The parameter identification unit calculates the following equations (1) to (7) for each sampling time Ts to obtain the motor parameter Θ=[ra Ld Lq Φm] ^T (ra is the winding resistance, Ld is the d-axis inductance, and Lq is the q-axis inductance, and Φm is the magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Ld' _n Lq _{' n} Φm' _n ] ^T , H=diag ( _hr , h _Ld , h _Lq , h _Φ ): constant (≧0), K= diag ( K _d , K _q ): constant (<0), H is a weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d-axis current, I _q is q-axis current, ω is angular frequency, e _dq is Error with respect to the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and the variable with ' indicates the estimated value).

The temperature estimating device for a PM motor according to claim 5 has the following features in claim 1:
The parameter identification unit calculates the following equations (10), (2), (3), (4), (5), (11), and (7) for each sampling time Ts. It is characterized by estimating the motor parameter Θ=[ra Ld Lq Φm] ^T (ra is the winding resistance, Ld is the d-axis inductance, Lq is the q-axis inductance, and Φm is the magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Ld' _n Lq _{' n} Φm' _n ] ^T , H=diag ( _hr , h _Ld , h _Lq , h _Φ ): constant (≧0), K= diag ( K _d , K _q ): constant (<0), H is a weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d-axis current, I _q is q-axis current, ω is angular frequency, e _dq is Error with respect to the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and the variable with ' indicates the estimated value).

The temperature estimating device for a PM motor according to claim 6 has the following features:
The parameter identification section calculates the following equations (12) to (16) at each sampling time Ts using the inductance value measured in advance, and calculates the motor parameter Θ=[ra Φm] ^T (ra is the winding resistance, Φm is characterized by estimating the magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T , H = diag ( _hr , h _Φ ): constant ( _hr , h _Φ ≧0), K = diag (K _d , K _q ): Constant (K _d , K _q <0), L _dqn = diag (L _dn , L _qn ): read from the table of inductance for current, H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d axis current, _Iq is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and variables with ' are estimated values. ).

The temperature estimating device for a PM motor according to claim 7 has the following features:
The parameter identification section calculates the following equations (17), (13), (14), (18), and (16) at each sampling time Ts using the inductance value measured in advance to determine the motor parameters. It is characterized by estimating Θ=[ra Φm] ^T (ra is the winding resistance and Φm is the magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T , H = diag ( _hr , h _Φ ): constant ( _hr , h _Φ ≧0), K = diag (K _d , K _q ): Constant (K _d , K _q <0), L _dqn = diag (L _dn , L _qn ): read from the table of inductance for current, H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d axis current, _Iq is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and variables with ' are estimated values. ).

(1) According to the invention described in claims 1 to 7, the resistance temperature and magnet temperature of the PM motor can be easily estimated without adding a torque sensor, a voltage sensor, or the like.
(2) According to the invention described in claim 1, motor parameters (winding resistance, d-axis inductance, q-axis inductance, magnet magnetic flux) can be estimated.
(3) According to the invention described in claim 2, motor parameters (winding resistance, magnet magnetic flux) can be estimated.
(4) According to the invention described in claims 4 to 7, it is possible to estimate temperature with high accuracy even if there is an error in inductance.
(5) According to the invention described in claims 6 and 7, since the inductance value measured in advance is used, the amount of calculation can be reduced compared to claims 4 and 5, and the parameters to be set can be reduced. is reduced from 4 parameters to 2 parameters, making adjustment easier.
(6) According to the inventions described in claims 5 and 7, since the model current (estimated current) is used instead of the measured current for model calculation, compared to claims 4 and 6, the current error is reduced. Since the change becomes larger, the responsiveness of identification improves.
(7) According to the inventions recited in claims 4 and 6, since the measured current is used in addition to the differential term of model calculation, the change in current error becomes gentler compared to claims 5 and 7. , the stability of identification is improved.

FIG. 1 is a configuration diagram of a temperature estimation device according to an embodiment of the present invention. FIG. 2 is a configuration diagram of a parameter identification section in Embodiment 1 of the present invention. FIG. 3 is a configuration diagram of a temperature calculation section in Embodiment 1 of the present invention. FIG. 2 is a configuration diagram of a parameter identification section in Example 2 of the present invention. FIG. 3 is a configuration diagram of a parameter identification section in Example 3 of the present invention. FIG. 4 is a configuration diagram of a parameter identification section in Embodiment 4 of the present invention. Graph showing temperature estimation simulation results. Graph showing motor parameter identification simulation results.

Embodiments of the present invention will be described below with reference to the drawings, but the present invention is not limited to the following embodiments. FIG. 1 shows a temperature estimating device according to this embodiment, and the magnet temperature of the PM motor of the present invention is estimated by the configuration shown in FIG.

In FIG. 1, 1 is a motor driven by an inverter 2 (IPMSM: Interior Permanent Magnet Synchronous Motor), 3 is a controller that gives a voltage command to the inverter 2, 4 is a motor current detector that detects the current flowing through the motor 1, and 5 6 is a motor rotation detector that detects the rotation of the motor 1, and 6 is a temperature estimator that estimates the resistance temperature of the stator winding of the motor 1 and the magnet temperature.

31 in the controller 3 is a multiplier that multiplies the detected rotation angle detected by the motor rotation detector 5 by the number of motor pole pairs to convert it into an electrical angle, and 32 differentiates the electrical angle output from the multiplier 31. This is a differentiator that calculates the angular frequency ω.

33 is a three-phase-dq converter that converts the three-phase motor detection current detected by the motor current detector 4 into a current in the dq coordinate system using the electrical angle output from the multiplier 31; It is.

34 controls the current so that the current (motor detection current) in the dq coordinate system coordinate-converted by the 3-phase-dq converter 33 becomes equal to the set current command, and converts the voltage command in the dq coordinate system. This is a current controller that outputs .

35 is a dq-3 phase converter which converts the voltage command in the dq coordinate system outputted from the current controller 34 into a three-phase voltage command using the electrical angle outputted from the multiplier 31 and supplies it to the inverter 2. It is a converter. The inverter 2 applies a voltage to the motor 1 according to the voltage command.

60 in the temperature estimator 6 is a current in a dq coordinate system coordinate-transformed by a three-phase-dq converter 33, an angular frequency ω determined by a differentiator 32, and a dq output from a current controller 34. This is a parameter identification unit that receives a voltage command in a coordinate system as an input and identifies motor parameters including stator winding resistance, magnet magnetic flux, inductance, etc. of the motor 1.

This parameter identification unit 60 provides the designed motor model 61 with the current (output current of the 3-phase-dq converter 33 (motor detection current)), voltage (voltage command output from the current controller 34), and angular frequency ( The model current of the motor model 61 is calculated by inputting the output of the differentiator 32 (the output of the differentiator 32), the difference between the calculated model current and the input current (motor detection current) is determined by the subtracter 62, and the difference is output from the subtracter 62. The stator winding resistance, magnet magnetic flux, and inductance of the motor 1 are identified by adjusting the winding resistance, magnet magnetic flux, and inductance of the motor model 61 using the parameter adjuster 63 so as to eliminate current errors.

68 is a temperature calculation unit that estimates the stator winding resistance temperature and magnet temperature from the stator winding resistance and magnet magnetic flux identified by the parameter identification unit 60.

In the first embodiment, the parameter identification _unit 60 in _{FIG .} 1 performs the IPMSM parameter estimation calculation shown in ^FIG _. Using I _d I _q ] ^T (voltage and current are values in the dq coordinate system), the following equations (1) to (7) are calculated. Here, the subscript n indicates the passage of time, and calculation is performed every sampling time _Ts . Furthermore, variables marked with '' indicate estimated values.

However, Θ _{' n} = [ra _{' n} Ld _{' n} Lq _{' n} Φm' _n ] ^T ,
H=diag( _hr , _hLd , _hLq , _hΦ ): constant (≧0),
K = diag (K _d , K _q ): constant (<0),
H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is the d-axis current, I _q is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the d-q-axis current, V _dq is the dq This is the shaft voltage command.

Note that 64, 65, and 66 in FIG. 2 are delay devices that respectively delay the estimated dq-axis current I' _dqn , the estimated parameter value Θ _{' n} , and the dq-axis current I _dqn , and the output side is equipped with a signal from one sampling time before. The estimated dq-axis current I _{' dqn-1} , the estimated parameter value Θ _{' n-1} one sampling time ago, and the dq-axis current I _dqn-1 one sampling time ago are output, respectively.

By repeating equations (1) to (7) every Ts, the motor parameter Θ=[r _a L _d L _q Φ _m ] ^T can be estimated.

The motor parameters Θ are winding resistance r _a , d-axis inductance L _d , q-axis inductance L _q , and magnet magnetic flux Φ _m .

As shown in FIG. 3, the temperature calculation section 68 calculates the stator winding using the following equations (8) and (9) from the winding resistance r' _an and the magnet magnetic flux Φ' _mn estimated by the motor parameter identification section 60. Calculate the resistance temperature _T'rn and magnet temperature _T'mn .

Here, r _a0 and Φ _m0 are the values of the winding resistance r _a and magnet magnetic flux Φ _m at temperature T ₀ , and α and β are the temperature coefficients of the winding resistance and magnet magnetic flux at temperature T ₀ . be. These values can be obtained from the motor manufacturer or determined through preliminary testing.

Next, the estimation calculations (formulas (1) to (9)) shown in FIGS. 2 and 3 performed by the temperature estimator 6 will be supplementarily explained.

The motor parameters (magnet magnetic flux, stator winding resistance, inductance) are estimated with the current, voltage, and angular velocity of the IPMSM known.

The IPM equation is as follows (expressed on d-q axes).

Then, equation (20) is obtained.

In the first embodiment, the following model calculation is performed for parameter estimation.

here,

Θ´＝[r _a ´ L _d ´ L _q ´ Φ _m ´] ^T
H=diag( _hr , _hLd , _hLq , _hΦ ).

<Design of P and Q using Lyapunov function>
An equation regarding the parameter error is derived from equations (20), (21), and (22), and if the function formed by the square of the parameter error is non-negative and the time differential is negative, the parameter error becomes 0.

<Error equation>
Substitute Θ'=Θ+e _Θ and L' _dq =L _dq +e _L into equation (20).

When formula (28) is selected as the Lyapunov function,

From formula (26) and formula (27)

therefore,

So,

Then,

, and the error e _Θ converges to 0.

<Temperature estimation>
The relationship between the resistance temperature T _r , the magnet temperature T _m , the resistance r _a , and the magnet magnetic flux Φ _m is expressed by the following equations (34) and (35).

r _a0 : resistance value [Ω] at T ₀ [°C], α: coefficient [1/°C]
Φ _m0 : Magnetic flux [Wb] at T ₀ [°C], β: Coefficient [1/°C]
Therefore, the temperature is estimated from the estimated resistance and magnetic flux as follows.

FIG. 7 shows temperature estimation simulation results when a current command is applied to the configuration of FIG. 1 so that the motor accelerates from 0 and reaches a constant speed. At this time, a constant load is applied, and the temperature is calculated from resistance loss and iron loss, and it is assumed that the resistance and magnetic flux change depending on the temperature.

In FIG. 7, the one-dot chain line shows the resistance temperature, the broken line shows the magnet temperature, the two-dot chain line shows the estimated resistance temperature, and the thin solid line shows the estimated magnet temperature. In the simulation shown in FIG. 7, although there is an error in estimating the resistance temperature in the first transient state, the temperature can almost be estimated. Because the transient response is slow, it is difficult to apply it to applications that require quick response, such as abnormality determination.

FIG. 8 shows the simulation results of motor parameter identification, and shows the ratio of the identified values of resistance and magnetic flux to each initial value when the simulation was performed under the same conditions as in FIG. 7. The inductance Lq was simulated by setting the initial value of the identified value to a value different from that of the motor Lq.

In FIG. 8, the one-dot chain line shows the resistance, the broken line shows the magnet magnetic flux, the two-dot chain line shows the estimated resistance, the thin solid line shows the estimated magnet magnetic flux, and the thick solid line shows the estimated q-axis inductance. According to FIG. 8, it can be seen that the motor parameters can be identified.

As described above, according to the first embodiment, the resistance temperature and magnet temperature of the PM motor can be easily estimated without adding a torque sensor, a voltage sensor, or the like. Further, even if there is an error in the inductance, accurate temperature estimation is possible. Furthermore, since the measured current is used in addition to the differential term in the model calculation, the change in current error becomes gradual, improving the stability of identification.

In the second embodiment, the parameter identification unit 60 in FIG. 1 performs the IPMSM parameter estimation calculation shown in FIG. 4, and the same parts as in the first embodiment (FIG. 2) are indicated by the same symbols.

The difference from Example 1 is that the measured current in equation (1) is replaced with the estimated current in equation (10), and accordingly, equation (6) is changed to equation (11) for calculation. The other parts operate in the same manner as in the first embodiment.

That is, the parameter identification unit 60 of the second embodiment calculates the angular frequency ω, the voltage command V _dq = [V _d V _q ] ^T , and the current I _dq = [I _d I _q ] ^T (voltage and current are expressed in the dq coordinate system. Calculate the following equations (10), (2), (3), (4), (5), (11), and (7) for each sampling time Ts using Then, the motor parameter Θ=[ra Ld Lq Φm] ^T (ra is the winding resistance, Ld is the d-axis inductance, Lq is the q-axis inductance, and Φm is the magnet magnetic flux) is estimated.

However, Θ _{' n} = [ra _{' n} Ld _{' n} Lq _{' n} Φm' _n ] ^T ,
H=diag( _hr , _hLd , _hLq , _hΦ ): constant (≧0),
K = diag (K _d , K _q ): constant (<0),
H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is the d-axis current, I _q is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the d-q-axis current, V _dq is the dq In the axis voltage command, the subscript n indicates the passage of time, and the variables with '' indicate estimated values.

Next, the estimation calculation (Equation 10), Equation (2), Equation (3), Equation (4), Equation (5), Equation (11), Equation (7)) of FIG. 4 is performed by the parameter identification unit 60. Here is a supplementary explanation.

The formula for parameter estimation is

as

Perform the calculation.

here

So,

Therefore, it can be considered in the same way as in the first embodiment. Therefore,

Parameters can be identified by calculating .

As described above, according to the second embodiment, the resistance temperature and magnet temperature of the PM motor can be easily estimated without adding a torque sensor, a voltage sensor, or the like. Further, even if there is an error in the inductance, accurate temperature estimation is possible. Furthermore, since the model current (estimated current) is used in the model calculation instead of the measured current, the change in current error becomes larger compared to Example 1, so the responsiveness of identification is improved.

In the third embodiment, the parameter identification unit 60 in FIG. 1 performs the IPMSM parameter estimation calculation shown in FIG. 5, and the same parts as in the first embodiment (FIG. 2) are indicated by the same reference numerals.

The difference from Embodiment 1 is that the inductance is measured in advance and stored in a table 67 of inductance with respect to current, and the winding resistance and magnet magnetic flux are identified.Other parts operate in the same manner as in FIG. do.

That is, the parameter identification unit 60 of the third embodiment calculates the angular frequency ω, the voltage command V _dq = [V _d V _q ] ^T , and the current I _dq = [I _d I _q ] ^T (voltage and current are expressed in the dq coordinate system. Equations (12) to (16) below are calculated for each sampling time ^Ts using .

However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T ,
H=diag( _hr , _hΦ ): constant ( _hr , _hΦ ≧0),
K=diag(K _d , K _q ): constant (K _d , K _q <0),
L _dqn = diag (L _dn , L _qn ): read from the table 67 of inductance with respect to current,
H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is the d-axis current, I _q is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the d-q-axis current, V _dq is the dq In the axis voltage command, the subscript n indicates the passage of time, and the variables with '' indicate estimated values.

The motor parameters are estimated by repeating the calculations of equations (12) to (16) every T _s , and based on the results, the temperature calculation section 68 calculates the resistance temperature T _{' rn} and the magnet temperature T as in the first embodiment. ' Calculate _mn .

Next, a supplementary explanation will be given of the estimation calculations (Equations (12) to (16)) shown in FIG. 5 performed by the parameter identification unit 60.

In equation (19), assuming that the inductance value is known, the equation for parameter estimation is written as

shall be.

<Design of P and Q using Lyapunov function>
Error equation Substitute I' _dq = I _dq + e _dq and Θ' = Θ' + e _Θ into equation (44)

shall be.

As described above, according to the third embodiment, the resistance temperature and magnet temperature of the PM motor can be easily estimated without adding a torque sensor, a voltage sensor, or the like. Further, even if there is an error in the inductance, accurate temperature estimation is possible. In addition, since the inductance value measured in advance is used, the amount of calculation can be reduced compared to Embodiment 1 and Embodiment 2, and the number of parameters to be set is reduced from 4 to 2, making adjustment easier. Become. Furthermore, since the measured current is used in addition to the differential term in the model calculation, the change in current error becomes gentler than in Example 2, and the stability of identification is improved.

In the fourth embodiment, the parameter identification unit 60 in FIG. 1 performs the IPMSM parameter estimation calculation shown in FIG. 6, and the same parts as in the third embodiment (FIG. 5) are indicated by the same symbols.

The difference from Example 3 is that the measured current in equation (12) is replaced with the estimated current in equation (17), and accordingly, equation (15) is changed to equation (18) for calculation. The other parts operate in the same manner as in the first embodiment.

That is, the parameter identification unit 60 of the fourth embodiment calculates the angular frequency ω, the voltage command V _dq = [V _d V _q ] ^T , and the current I _dq = [I _d I _q ] ^T (voltage and current are expressed in the dq coordinate system. value), calculate the following equations (17), equation (13), equation (14), equation (18), and equation (16) for each sampling time Ts to obtain the motor parameter Θ=[ra Φm] ^T (ra is the winding resistance, Φm is the magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T ,
H=diag( _hr , _hΦ ): constant ( _hr , _hΦ ≧0),
K=diag(K _d , K _q ): constant (K _d , K _q <0),
L _dqn = diag (L _dn , L _qn ): read from the table 67 of inductance with respect to current,
H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is the d-axis current, I _q is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the d-q-axis current, V _dq is the dq In the axis voltage command, the subscript n indicates the passage of time, and the variables with '' indicate estimated values.

The motor parameters are estimated by repeating the calculations of Equations (17), (13), Equations (14), Equations (18), and Equations (16) every T _s , and from the results, the temperature calculation unit 68 calculates the motor parameters. Similarly to the first embodiment, the resistance temperature _T'rn and the magnet temperature _T'mn are calculated.

Next, the estimation calculations (Equation (17), Equation (13), Equation (14), Equation (18), and Equation (16)) shown in FIG. 6 performed by the parameter identification unit 60 will be supplementally explained.

The current in equation (43) is changed as follows.

shall be.

<Design of P and Q using Lyapunov function>

Substitute I' _dq = I _dq + e _dq and Θ' = Θ' + e _Θ into equation (44)

here

It is said that

Since equation (53) has the same form as equation (45),

shall be.

As described above, according to the fourth embodiment, the resistance temperature and magnet temperature of the PM motor can be easily estimated without adding a torque sensor, a voltage sensor, or the like. Further, even if there is an error in the inductance, accurate temperature estimation is possible. In addition, since the model current (estimated current) is used for model calculation instead of the measured current, the change in current error becomes larger compared to Examples 1 and 3, improving the responsiveness of identification. do. In addition, since the inductance value measured in advance is used, the amount of calculation can be reduced compared to Embodiment 1 and Embodiment 2, and the number of parameters to be set is reduced from 4 to 2, making adjustment easier. Become.

1... Motor 2... Inverter 3... Controller 4... Motor current detector 5... Motor rotation detector 6... Temperature estimator 31... Multiplier 32... Differentiator 33... 3-phase-dq converter 34... Current controller 35... dq -3-phase converter 60...Parameter identification unit 61...Current controller 62...Subtractor 63...Parameter adjuster 64, 65, 66...Delay unit 67...Table 68...Temperature calculation unit

Claims (7)

Current control is performed so that the motor detection current obtained by detecting the current flowing through the PM motor driven by the inverter, the angular frequency obtained by detecting the rotation speed of the motor, and the motor detection current become equal to the current command. Using the obtained voltage command as input, calculate the current of the motor model using the motor model, and adjust the winding resistance and magnet magnetic flux of the motor model so as to eliminate the difference between the calculated motor model current and the motor detection current. , a parameter identification unit that identifies motor parameters including motor winding resistance, magnet magnetic flux, and inductance by adjusting inductance;
A temperature estimation device for a PM motor, comprising: a temperature calculation section that estimates a resistance temperature and a magnet temperature from the winding resistance and magnet magnetic flux identified by the parameter identification section. Current control is performed so that the motor detection current obtained by detecting the current flowing through the PM motor driven by the inverter, the angular frequency obtained by detecting the rotation speed of the motor, and the motor detection current become equal to the current command. Using the obtained voltage command as input, calculate the current of the motor model using the motor model, and adjust the winding resistance and magnet magnetic flux of the motor model so as to eliminate the difference between the calculated motor model current and the motor detection current. a parameter identification unit that identifies motor parameters including motor winding resistance and magnet magnetic flux by adjusting the
A temperature estimation device for a PM motor, comprising: a temperature calculation section that estimates a resistance temperature and a magnet temperature from the winding resistance and magnet magnetic flux identified by the parameter identification section. The inverter drives the PM motor by vector control, and the vector control uses voltage and current in a dq coordinate system synchronized with the rotor of the PM motor. PM motor temperature estimation device. The parameter identification unit calculates the following equations (1) to (7) for each sampling time Ts to obtain the motor parameter Θ=[ra Ld Lq Φm] ^T (ra is the winding resistance, Ld is the d-axis inductance, and Lq 2. The PM motor temperature estimating device according to claim 1, wherein: q-axis inductance, Φm: magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Ld' _n Lq _{' n} Φm' _n ] ^T , H=diag ( _hr , h _Ld , h _Lq , h _Φ ): constant (≧0), K= diag ( K _d , K _q ): constant (<0), H is a weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d-axis current, I _q is q-axis current, ω is angular frequency, e _dq is Error in the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and the variables with ' indicate the estimated value) The parameter identification unit calculates the following equations (10), (2), (3), (4), (5), (11), and (7) for each sampling time Ts. The PM according to claim 1, characterized in that the motor parameter Θ=[ra Ld Lq Φm] ^T (ra is winding resistance, Ld is d-axis inductance, Lq is q-axis inductance, and Φm is magnet magnetic flux) is estimated. Motor temperature estimation device.

(However, Θ _{' n} = [ra _{' n} Ld' _n Lq _{' n} Φm' _n ] ^T , H=diag ( _hr , h _Ld , h _Lq , h _Φ ): constant (≧0), K= diag ( K _d , K _q ): constant (<0), H is a weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d-axis current, I _q is q-axis current, ω is angular frequency, e _dq is Error in the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and the variables with ' indicate the estimated value) The parameter identification section calculates the following equations (12) to (16) at each sampling time Ts using the inductance value measured in advance, and calculates the motor parameter Θ=[ra Φm] ^T (ra is the winding resistance, Φm 3. The PM motor temperature estimating device according to claim 2, wherein the PM motor temperature estimating device estimates a magnetic flux of a magnet.

(However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T , H = diag ( _hr , h _Φ ): constant ( _hr , h _Φ ≧0), K = diag (K _d , K _q ): Constant (K _d , K _q <0), L _dqn = diag (L _dn , L _qn ): read from the table of inductance for current, H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d axis current, _Iq is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and variables with ' are estimated values. ) The parameter identification section calculates the following equations (17), (13), (14), (18), and (16) at each sampling time Ts using the inductance value measured in advance to determine the motor parameters. 3. The PM motor temperature estimating device according to claim 2, wherein Θ=[ra Φm] ^T (ra is winding resistance and Φm is magnet magnetic flux).

(However, Θ _{' n} = [ra _{' n} Φm _{' n} ] ^T , H = diag ( _hr , h _Φ ): constant ( _hr , h _Φ ≧0), K = diag (K _d , K _q ): Constant (K _d , K _q <0), L _dqn = diag (L _dn , L _qn ): read from the table of inductance for current, H is the weighting coefficient, diag is a function that outputs a diagonal matrix, I _d is d axis current, _Iq is the q-axis current, ω is the angular frequency, e _dq is the error for the estimated value of the dq-axis current, V _dq is the dq-axis voltage command, the subscript n indicates the passage of time, and variables with ' are estimated values. )

Images