JP2022184581A - Magnet temperature estimation method for motor - Google Patents

Abstract

To estimate a magnet temperature with high accuracy regardless of an operation state of a motor.SOLUTION: Both of a first magnet temperature estimation processing for estimating a temperature of a magnet based on an actual machine of a motor 2 by using a state amount inputted to the motor 2 from an inverter 3 and a second magnet temperature estimation processing for estimating the temperature of the magnet based on a heat model of a motor 3 are executed. An error feedback is executed, and according to a difference between a first estimation value and a second estimation value, the state amount used for the second magnet temperature estimation processing is corrected by a predetermined correction coefficient K. A noise superimposed to the state amount is estimated, and according to the noise, the correction coefficient K is changed.SELECTED DRAWING: Figure 14

Description

The disclosed technique relates to a magnet temperature estimation method for a motor.

In recent years, the number of vehicles that run on electric power, such as hybrid vehicles and electric vehicles, is increasing. Usually, such a vehicle is equipped with a permanent magnet synchronous motor (simply referred to as a motor) as a drive source. In the motor, a permanent magnet (simply called a magnet) is installed in the rotor. An alternating current controlled by an inverter is applied to the stator of the motor. In doing so, the rotor rotates under the action of the changing magnetic field, driving the motor.

When the motor drives, the magnet heats up. Therefore, when the motor is continuously driven for a long time, the magnet becomes hot. In contrast, the maximum magnetic flux of a magnet is temperature dependent. When the temperature of the magnet exceeds a predetermined upper limit, the magnetic force of the magnet rapidly drops, and the dropped magnetic force does not recover. That is, the magnet is irreversibly demagnetized.

Therefore, it is necessary to strictly control the temperature of the magnet so as not to exceed the upper limit while the motor is being driven. To do that, the temperature of the magnet must be measured, while the rotor is rotating. It is difficult to directly measure the temperature of the magnet using a temperature sensor or the like.

Therefore, conventionally, the temperature of the magnet is estimated by an indirect method based on the temperature of the stator, the induced voltage, and the like (for example, Patent Document 1).

Moreover, there is Patent Document 2 as a prior art related to an application example of the technique disclosed here. It discloses a technique for estimating the temperature of the motor, specifically the transient temperature of the coil.

WO 2005/010001 also discloses a block diagram of state feedback control, in which a full state observer is used. Then, the error between the actual thermistor temperature obtained by actually measuring the temperature of the coil of the actual motor, which is the system, and the estimated thermistor temperature output from the simulation model, which is the observer, is multiplied by the feedback gain to input the simulation model. is added to correct the estimated thermistor temperature.

A predetermined value is used for the feedback gain (paragraph 0040).

WO2014-057558 JP 2016-082698 A

The estimated value may have a large error. Therefore, when controlling the temperature of the magnet based on the estimated value, the error must be taken into account, so the temperature of the magnet must be controlled at a temperature excessively lower than the upper limit. As a result, at present, the motor cannot fully demonstrate its performance. There is a need to improve the accuracy of magnet temperature estimation.

In estimating the temperature of the magnet, it is often the case that the state quantity such as the value of the electric current supplied to the motor is actually measured and the temperature is estimated based on the measured value. Since the actual measurement requires high sensitivity and high accuracy, various noises are superimposed on the actually measured state quantity. Processing noise with a low-pass filter (LPF) is also conceivable, but the processing takes a long time to some extent. Therefore, when estimating the magnet temperature in a very short time such as millisecond level, the LPF is not suitable for noise processing.

On the other hand, if the state feedback control as disclosed in Patent Document 2 is used, noise can be processed even in such an extremely short period of time. However, in cases such as motors mounted on vehicles, where there are a wide variety of noises and can include large noises, even if state feedback control is used, stable control can be achieved while maintaining both responsiveness and accuracy. is difficult to achieve.

The technology disclosed herein aims to realize a method for estimating the magnet temperature of a motor with excellent estimation accuracy and responsiveness.

The disclosed technology is a method for estimating the magnet temperature of a motor in which magnets that make up magnetic poles are installed in a rotor, and the motor rotates by energizing a plurality of stator coils with an alternating current formed by controlling an inverter. Regarding. Then, the magnet temperature estimation method executes each process shown below.

a first magnet temperature estimation process for estimating the temperature of the magnet based on the actual motor using the state quantity input from the inverter to the motor; and and a second magnet temperature estimation process for estimating the temperature. According to the difference between the first estimated value estimated by the first magnet temperature estimation process and the second estimated value estimated by the second magnet temperature estimation process, in the second magnet temperature estimation process Error feedback is performed to correct the state quantity used by a predetermined correction factor. Then, the noise superimposed on the state quantity is predicted, and the correction coefficient is changed according to the noise.

That is, according to this magnet temperature estimation method, the process of estimating the temperature of the magnet based on the actual motor (first magnet temperature estimation process) and the process of estimating the temperature of the magnet based on the thermal model of the motor ( second magnet temperature estimation process) are executed. Each has a different presumed mechanism.

Then, by executing error feedback, the second magnet temperature estimation process is performed according to the difference (estimation error) between the two estimated values (first estimated value and second estimated value) estimated by both processes. is corrected by a predetermined correction coefficient. By doing so, the estimation error can be converged. Since the estimated value can be asymptotic to the true value, highly accurate estimation can be performed. Moreover, according to this method, it is not necessary to obtain a large number of estimated values in order to improve the estimation accuracy, so estimation can be performed in a short period of time.

However, in the first magnet temperature estimation process, the magnet temperature is estimated based on the actual motor. Therefore, when large noise is superimposed on the state quantity, the estimated value deviates from the true value. As a result, the estimation error increases, and the magnitude of the correction coefficient becomes inappropriate, which may prevent the estimation error from quickly converging.

On the other hand, in this magnet temperature estimation method, noise superimposed on the state quantity is predicted, and the correction coefficient is changed according to the noise. That is, since the correction coefficient is predictively changed, even if a large amount of noise is superimposed on the state quantity, the correction coefficient can be changed according to the magnitude of the noise before error feedback is performed. As a result, error feedback can be appropriately performed. Even if a large amount of noise is superimposed on the state quantity, both convergence and stability can be achieved in an appropriate state. As a result, the magnet temperature can be quickly estimated with high accuracy.

In the magnet temperature estimation method, the state quantity may be at least one of a current and a voltage, and the correction coefficient may be changed so as to decrease as the state quantity increases.

Current and voltage are state quantities input from the inverter to the motor. Therefore, it is a direct factor affecting noise. As the current and voltage increase, the noise superimposed on them increases accordingly. As the current and voltage become smaller, the noise superimposed on them becomes smaller accordingly. That is, there is a positive correlation between the magnitude of current and voltage and the magnitude of noise superimposed thereon.

Therefore, if the correction coefficient is changed so as to decrease as the state quantity of the current and voltage increases, the correction coefficient can be changed according to the magnitude of the current and voltage on which noise is superimposed. As a result, error feedback can be appropriately performed. As a result, the magnet temperature can be quickly estimated with high accuracy.

The magnet temperature estimating method further includes, as the first temperature estimating process, momentarily interrupting the energization of the coil during operation of the motor, a current value at momentary interruption, which is a current value at the time of momentary interruption, and A transient coil voltage value, which is the voltage value of the coil during the momentary interruption transient, is obtained, and a predetermined formula formulated based on the voltage change of the coil after the momentary interruption, and the momentary interruption current value and the transient coil voltage value, the induced voltage may be estimated, and the temperature of the magnet may be estimated based on the calculated induced voltage.

That is, according to this magnet temperature estimation method, a momentary stop (instantaneous interruption or momentary interruption) is performed in the first temperature estimation process for estimating the temperature of the magnet based on the actual motor. Therefore, it hardly affects the torque output by the motor. If it is a momentary interruption, it can be performed regardless of the operating state of the motor. Since it can be performed not only during the deceleration period of the vehicle but also during the acceleration period, it is excellent in responsiveness.

Then, at the time of the instantaneous interruption, the current value at the time of instantaneous interruption and the coil voltage value at the time of transition are acquired, and the induced voltage is calculated using these values and a predetermined formula. Since the formula is formulated based on the voltage change of the coil after the momentary interruption, the induced voltage after the transition can be calculated by using the formula. The induced voltage can be calculated by actually measuring the instantaneous current value and the transient coil voltage value and substituting these values into the formula. After that, the temperature of the magnet should be estimated based on the calculated induced voltage.

Therefore, the magnet temperature can be estimated with high accuracy regardless of the operating state of the motor. As a result, appropriate temperature control of the magnet is facilitated, so that the performance of the motor can be sufficiently exhibited, and consumption of fuel and/or battery can be suppressed.

In the magnet temperature estimation method, the noise may include inherent noise caused by the inverter itself, and the magnitude of the correction coefficient may be changed according to the inherent noise.

Various electronic components are densely packed inside the inverter, and a large current flows. Therefore, highly sensitive actual measurements inside or near the inverter are likely to be affected by them. Therefore, the inherent noise tends to become large noise. Therefore, if the magnitude of the correction coefficient is changed according to the inherent noise, the effect can be suppressed, so the magnet temperature can be estimated quickly and accurately.

The magnet temperature estimating method also includes external noise generated due to an electrical component other than the inverter mounted on the electric vehicle, wherein the motor is mounted as a drive source on the electric vehicle. The magnitude of the correction coefficient may be changed according to external noise.

Such external noise can also become large noise. Therefore, if the magnitude of the correction coefficient is changed according to the external noise, the influence can be suppressed, so the magnet temperature can be quickly estimated with high accuracy.

In the magnet temperature estimation method, the motor is mounted as a drive source on an electric vehicle, includes specific environmental noise generated due to a specific environment in which the electric vehicle travels, and to change the magnitude of the correction coefficient.

Such specific environmental noise can also become large noise. Therefore, by changing the magnitude of the correction coefficient according to the specific environmental noise, the influence can be suppressed, so the magnet temperature can be quickly and highly accurately estimated.

According to the disclosed technology, highly accurate estimation of the magnet temperature is possible. As a result, appropriate temperature control of the magnet is facilitated, so that the performance of the motor can be sufficiently exhibited, and consumption of fuel and/or battery can be suppressed.

1 is a simplified diagram of a motor system suitable for application of the disclosed technology; FIG. 1 is a diagram showing a simplified electric circuit of a motor system; FIG. The upper diagram shows the operating state of the motor, and the lower diagram shows the stopped state of the motor. FIG. 10 is a diagram for explaining a current change in off control of a switching element; 5 is a graph for explaining changes in coil voltage immediately after the motor stops; FIG. 10 is a diagram showing an equivalent circuit model of the motor system at the time of momentary blackout transient; It is a list of characters used in mathematical formulas and the like. It is a figure showing Numerical formula (1) corresponding to description. It is a figure for demonstrating Numerical formula (2). It is a figure showing Formula (3) and (4) corresponding to description. FIG. 10 is a diagram showing formulas (5) to (7) corresponding to the description; FIG. 11 is a diagram showing formulas (8) to (11) corresponding to the description; 4 is a flowchart of processing for estimating magnet temperature. FIG. 2 is a block diagram showing a conventional observer model (upper diagram) and a modified observer model (lower diagram); 1 is a schematic configuration diagram of a magnet temperature estimation system to which a magnet temperature estimation method of an application example is applied; FIG. FIG. 2 is a diagram simply showing an example of a thermal model; FIG. 4 is a diagram for explaining a correspondence relationship between feedback gain K and current and voltage; 7 is a flowchart of processing related to magnet temperature estimation in an application example. It is the graph which showed the outline of the demonstration test result.

Embodiments of the disclosed technology will be described in detail below with reference to the drawings. However, the following description is merely exemplary in nature. For convenience, mathematical formulas relating to the disclosed technology are presented both in the specification and in the drawings. Common letters are used in the formulas (see table shown in FIG. 6).

<Motor system>
FIG. 1 shows a simplified diagram of a motor system 1 suitable for application of the disclosed technology. This motor system 1 is installed in a vehicle (electric vehicle) that can run using electric power, such as a hybrid vehicle and an electric vehicle. A motor system 1 includes a driving motor 2, an inverter 3, a motor control unit (MCU 4), and the like.

The motor 2 is a permanent magnet synchronous motor. The motor 2 has a rotor 20 and a stator 21 . The rotor 20 is arranged inside the stator 21 . The rotor 20 may be arranged outside the stator 21, but this motor 2 is of the inner rotor type. A shaft 22 is fixed at the center of the rotor 20 . The shaft 22 is rotatably supported by a motor case housing the rotor 20 and the stator 21 . Rotational power of the motor 2 is output via the shaft 22 . That is, the motor 2 drives. The vehicle is thereby driven.

Permanent magnets 23 (also simply referred to as magnets 23 ) are installed on the outer peripheral portion of the rotor 20 . The magnets 23 constitute magnetic poles of the rotor 20 . That is, the magnets 23 are arranged on the outer peripheral portion of the rotor 20 so that the S poles and the N poles are alternately arranged at regular intervals in the circumferential direction. The number, arrangement, shape, etc. of the magnets 23 are set according to the specifications of the motor 2 . For example, a plurality of plate-shaped magnets 23 may be installed, or one annular magnet 23 having a plurality of magnetic poles may be installed.

The stator 21 is arranged around the rotor 20 with a small gap (gap 24) therebetween. The stator 21 has a stator core 21 a and multiple coils 25 . A plurality of coils 25 are formed by winding electric wires in a predetermined order around a plurality of teeth extending radially from the inside of the stator core 21a.

The shape of the stator core 21a, the number and arrangement of the coils 25, etc. are set according to the specifications of the motor 2. FIG. In the case of this motor 2, these coils 25 constitute three coil groups 25U, 25V and 25W consisting of U-phase, V-phase and W-phase. Each of these coil groups 25U, 25V, 25W is connected to inverter 3 via connection cable 26 .

Inverter 3 is connected to DC power supply 5 via power supply cable 30 . Specific examples of the DC power supply 5 include a low voltage battery with a rated voltage of 50 V or less and a high voltage battery with a rated voltage of 300 V or more. DC power supply 5 supplies power to inverter 3 .

Inverter 3 is also connected to MCU 4 via harness 40 . The MCU 4 is composed of hardware such as a processor, memory and interface, and software such as a control program. By combining these hardware and software, the MCU 4 is provided with an inverter control section 41, a magnet temperature estimation section 42, and a database 43 as functional configurations.

The inverter control unit 41 drives the motor 2 by controlling the inverter 3 . The magnet temperature estimator 42 estimates the temperature of the magnet 23 using a predetermined formula (circuit equation, etc.) to be described later. The database 43 stores data used for controlling the inverter 3 and estimating the temperature of the magnet 23 . There are various data formats, and the data is stored in the database 43 in formats such as numerical values, formulas, maps, tables, etc., depending on the application. The inverter control section 41 and the magnet temperature estimation section 42 use these data as necessary.

FIG. 2 shows a simplified electric circuit of the motor system 1. As shown in FIG. The inverter 3 incorporates a known inverter circuit 31 . The inverter circuit 31 of the inverter 3 converts the power of the DC power supply 5 into three-phase AC current, and energizes each of the coil groups 25U, 25V, and 25W of the motor 2 . Thereby, a periodically changing magnetic field is formed between the magnetic poles of rotor 20 and each of coil groups 25U, 25V, and 25W of stator 21 . As the magnetic field changes, the rotor 20 and shaft 22 rotate and the motor 2 drives.

In the DC power supply 5, R represents internal resistance and V represents DC voltage. The inverter circuit 31 incorporated in the inverter 3 has three arms 31a corresponding to the U, V, and W phases. These arms 31 a are connected to the high voltage (plus) side and the low voltage (minus) side of the DC power supply 5 via power supply cables 30 .

Two switching elements (an upstream switching element 33U and a downstream switching element 33L) are connected in series to each of these arms 31a. The switching elements 33U and 33L are, for example, semiconductor switches such as MOS-FETs and IGBTs. Two diodes 34 are also connected to each of the arms 31a in antiparallel with each of the switching elements 33U and 33L. These diodes 34 are so-called freewheeling diodes.

A connection cable 26 connected to each phase coil group 25U, 25V, 25W of the motor 2 is connected between each of the upstream switching element 33U and the downstream switching element 33L. In each of the coil groups 25U, 25V, 25W, r represents the internal resistance and L represents the inductance of the coil 25 of each phase. The inverter circuit 31 forms an alternating current by controlling the on/off of these switching elements 33U and 33L, and energizes the coil groups 25U, 25V and 25W of each phase of the motor 2 .

The upper diagram of FIG. 2 represents a predetermined state during operation of the motor 2 . Here, the U-phase upstream switching element 33U, the V-phase downstream switching element 33L, and the W-phase downstream switching element 33L are on-controlled. On the other hand, the U-phase downstream switching element 33L, the V-phase upstream switching element 33U, and the W-phase upstream switching element 33U are turned off.

As a result, currents are supplied to the coil groups 25U, 25V, and 25W of each phase, as indicated by dashed arrows. Current flows into the motor 2 through the U-phase connection cable 26, and current flows out of the motor 2 through the V-phase and W-phase connection cables 26 (the U-phase current is positive, and the V-phase and W-phase currents are negative). . The inverter control unit 41 performs on/off control of the switching elements 33U and 33L at predetermined timings. By doing so, the current flow of each phase is switched such that, for example, the V-phase current is positive and the W-phase and U-phase currents are negative. Alternating currents of different phases are supplied to the coil groups 25U, 25V, and 25W of each phase.

The lower diagram in FIG. 2 shows the state when the motor 2 is stopped from the state shown in the upper diagram by cutting off the energization. That is, the U-phase upstream switching element 33U, the V-phase downstream switching element 33L, and the W-phase downstream switching element 33L, which have been turned on, are turned off. At this time, an induced electric power is generated in the coil groups 25U, 25V, and 25W of each phase, and current (induced current) flows for a short time as indicated by dashed arrows.

When the switching elements 33U and 33L are controlled to be off, ideally, they are switched from the on state to the off state without a time lag. However, in practice there is a transient state between the on and off states. FIG. 3 shows changes in current before and after on/off control of the switching elements 33U and 33L. The off control is the timing of t0.

In ideal off control, the current becomes 0 at timing t0, as indicated by the dashed line. There are no transients. On the other hand, in the actual OFF control, as indicated by the solid line, the current does not become 0 at the timing t0, but becomes 0 at a predetermined timing after t0 (fall of the current). That is, the current flows for a certain period of time even after t0.

Therefore, in the case of actual OFF control, the current flow in the motor system 1 immediately after the motor 2 stops is not only the current path indicated by the broken arrow in the lower diagram of FIG. 2, but also the current indicated by the solid arrow. (This current is also called "leakage current").

<Estimation of magnet temperature>
When the motor 2 drives, the magnet 23 generates heat. Therefore, when the motor 2 is continuously driven for a long time, the magnet 23 becomes hot. On the other hand, the maximum magnetic flux of magnet 23 is temperature dependent. When the temperature of the magnet 23 (hereinafter also referred to as magnet temperature) rises, the magnetic flux density decreases. Then, when the magnet temperature exceeds a predetermined upper limit value, the magnetic force of the magnet 23 drops sharply, and the dropped magnetic force does not recover. That is, the magnet 23 is irreversibly demagnetized.

Therefore, it is necessary to strictly control the temperature of the magnet 23 so that the upper limit is not exceeded while the motor 2 is being driven. To do so, the magnet temperature must be measured, and the magnet 23 is installed on the rotating rotor 20 . Therefore, it is difficult to directly measure the magnet temperature. The magnet temperature can only be estimated by an indirect method.

Therefore, the MCU 4 is provided with a magnet temperature estimating section 42 . The magnet temperature estimator 42 estimates the magnet temperature based on the induced voltage (also referred to simply as induced voltage) caused by the magnetic flux of the magnet 23 . Since there is a correlation between the induced voltage and the magnet temperature, if the induced voltage (specifically, the maximum value of the induced voltage, etc.) can be detected with high accuracy, the magnet temperature can also be estimated with high accuracy.

However, the conventional induced voltage detection method requires stopping the motor 2 and waiting for a certain amount of time to pass. Therefore, there is a disadvantage of lacking responsiveness.

FIG. 4 illustrates voltage changes of the coil 25 immediately after the motor 2 is stopped, that is, immediately after the energization of the coil 25 is stopped. The motor 2 is stopped at the timing t0. The solid-line graph Gy represents the induced voltage, and the dashed-line graph Gz represents the terminal voltage applied to the coil 25 (the sum of the induced voltage and the back electromotive force).

As described above, when the motor 2 is stopped, an induced current flows and a back electromotive force is generated in the coil 25 . As a result, the coil voltage immediately after the motor 2 stops is the sum of the back electromotive force and the induced voltage. When time elapses and the induced current stops flowing and the back electromotive force becomes 0 (timing t1), the coil voltage becomes only the induced voltage, and the induced voltage can be detected.

That is, the conventional method for detecting the induced voltage detects the induced voltage at the timing after t1, so the operation of the motor 2 is restricted. For example, if the induced voltage is detected while the motor 2 is running, there will be a time during which torque cannot be output. As a result, a torque shock occurs and the vehicle cannot run smoothly. In order not to adversely affect the traveling of the vehicle, it can be used only when the time during which no torque is output can be ensured, for example, during the deceleration period of the vehicle. That is, the conventional method for detecting the induced voltage is limited by the operating state of the motor 2, and lacks responsiveness.

On the other hand, the magnet temperature estimator 42 momentarily stops (momentarily interrupts) the energization of the coil 25 by turning off the switching elements 33U and 33L, for example, between t0 and t2. The time during which the energization is stopped due to an instantaneous interruption is, for example, 10 milliseconds (ms) or less. 5 milliseconds or less is preferred. Then, the induced voltage is estimated using mathematical formulas such as a predetermined circuit equation newly formulated this time.

Specifically, the current value at the moment of momentary interruption (at or before or after t0) and the transient moment of momentary interruption (period from momentary interruption to steady state, timing from t0 to t1) obtained by actually measuring the voltage value of the coil 25 (coil voltage value during transition) at . Then, these instantaneous interruption current value and transient coil voltage value are substituted into a predetermined circuit equation (more specifically, an analytical solution equation derived from the circuit equation). Then, the induced voltage is estimated.

That is, the induced voltage is detected by a momentary stop (instantaneous interruption) that hardly affects the operation of the motor 2 . Since it is sufficient to momentarily stop the energization of the coil 25, the torque output by the motor 2 is hardly affected. Therefore, regardless of the operating state of the motor 2, the induced voltage can be detected. Since the induced voltage can be detected not only during the deceleration period of the vehicle but also during the acceleration period, it is excellent in responsiveness.

<Circuit equation>
(basic circuit equation)
FIG. 5 shows an equivalent circuit model of the motor system 1 during an instantaneous power failure transient. This equivalent circuit model corresponds to FIG. Note that common characters are used in the following mathematical formulas including FIG. 5 (see the table shown in FIG. 6).

That is, it represents the case where the U-phase current is positive and the V-phase and W-phase currents are negative. The equivalent circuit model in the case where the V-phase or W-phase is momentarily interrupted when the current is positive is simply replacing the portion corresponding to the U-phase with each phase. Therefore, for the sake of convenience, the following description assumes that the phase through which the positive current was conducted at the moment of momentary interruption (also referred to as the phase conducting current at momentary interruption) is the U phase, unless otherwise specified.

In this figure, the leakage currents mentioned above are also taken into account. The induced voltage Vemf of each phase is shown in a simplified manner. If the induced voltage of each phase is shown correctly, it will be as follows.

U-phase induced voltage: V _emf sin(ωt)
Induced voltage of V phase: V _emf sin (ωt-2π/3)
W-phase induced voltage: V _emf ·sin(ωt+2π/3).

Based on this equivalent circuit model, a new circuit equation (basic circuit equation) was established this time. FIG. 7 shows the basic circuit equation as Equation (1). This basic circuit equation expresses a transient state after an instantaneous interruption, and shows a basic circuit equation in which the U phase is the energized phase during an instantaneous interruption.

The solution of this basic circuit equation can be obtained analytically by Laplace transform. When the analytic solution formula obtained from the basic circuit equation is organized and simplified with respect to the induced voltage, it can be expressed by the following formula.

V _emf =Ka· _VP (t)+Kb·I ₀ +Kc General formula (1)
Here, V _P (t) is the transient coil voltage value of the energized phase during momentary interruption, and I ₀ is the current value during momentary interruption. Ka, Kb, and Kc are predetermined coefficients. Ka, Kb, and Kc are coefficients determined by the contents of the circuit equations and include variables.

That is, by obtaining the instantaneous current value and the transient coil voltage value, the induced voltage can be estimated using the basic circuit equation. Therefore, the basic circuit equations or formulas obtained from the basic circuit equations should be stored in the MCU4. The instantaneous power failure current value and the transient coil voltage value may be obtained by actually measuring them with the inverter 3 and outputting the data to the MCU 4 . The MCU 4 can then estimate the induced voltage.

For the basic circuit equation, instead of using the current value at the time of instantaneous interruption and the coil voltage value at the time of transient, as described later, the induced voltage is calculated using the slope of the coil voltage of the energized phase at the moment of instantaneous interruption. can be estimated.

<Simplification of basic circuit equation>
Estimation of the induced voltage using the basic circuit equation can estimate the induced voltage with high accuracy, but the computational load is large in the calculation of the analytical solution. Therefore, a high-performance processor is required, and even a high-performance processor is difficult to operate in a short time. Therefore, it is not efficient to use the basic circuit equation itself to estimate the induced voltage each time the magnet temperature is estimated.

On the other hand, in practice, there are many cases in which a certain amount of error in the estimation accuracy of the induced voltage is permissible. Also, depending on the operating state of the motor 2, the basic circuit equation may be simplified. Therefore, in such cases, it is preferable to use the simplified basic circuit equations described above. It is possible to reduce the burden of arithmetic processing and increase the processing speed. Therefore, next, a method for simplifying the basic circuit equation will be described.

(Data conversion of variable inclusion terms)
As shown in FIG. 7, the basic circuit equation has terms containing variables (variable-containing terms). For example, the induced current is variable because it changes during a power outage transient. The inductance varies depending on the rotation angle of the motor 2 and the temperature of the coil 25 as well. Therefore, when considering these, inductance is also a variable. Since the resistance of the switching elements 33U and 33L also changes with time, the resistance of the switching elements 33U and 33L is also a variable when considering this.

When calculating using the basic circuit equations, the more variables there are, the greater the computational load. Therefore, it is preferable to store in the database 43 data in advance for variables that can be replaced with numerical values by experiment or the like. The entire variable-containing term may be converted to data, or part of the variable-containing term may be converted to data. The data format may be appropriately selected according to its use.

(First simplified circuit equation)
The basic circuit equations are established considering the actual OFF control in which leakage current occurs. On the other hand, as shown in the upper part of FIG. 8, by assuming that the resistances Rl of the switching elements 33U and 33L are infinite, it is possible to express ideal off control in which no leakage current occurs. By doing so, the equivalent circuit model shown in FIG. 5 can be simplified. The middle part of FIG. 8 shows the simplified equivalent circuit model (simplified equivalent circuit model).

By formulating a circuit equation based on this simplified equivalent circuit model, a simplified circuit equation (first simplified circuit equation) that does not consider leakage current is obtained. The first simplified circuit equation is shown as Equation (2) in the lower part of FIG. Note that the resistance r of the coil 25 is not taken into consideration in Equation (2) for simplification (if it is taken into consideration, R should be replaced with R+2r).

(Constantization of inductance)
As described above, the inductance changes depending on the rotation angle of the motor 2, the temperature of the coil 25, and the like. However, if the degree of change can be practically ignored, the inductance of the coil 25 for each phase may be constant. Specifically, in the basic circuit equations or simplified circuit equations, the same constant value L is used as the inductances Lu, Lv, and Lw of the coils 25 of each phase, as shown in the upper part of FIG.

As an example, in the first simplified circuit equation, the equation obtained by making the inductance of the coil 25 of each phase constant is shown as equation (3) in FIG.

(Derivation of analytical solution)
Next, the derivation of the analytical solution will be described using the first simplified circuit equation shown in Equation (3).

The Laplace transform derives the current from the first simplified circuit equation shown in Equation (3). Then, the U-phase coil voltage VU(t) is obtained from the relationship v=L·di/dt. An analytical solution formula obtained thereby is shown as formula (4) in FIG.

In Equation (4), α is 2R/3L. t is the elapsed time after the momentary interruption. The transient coil voltage value can be calculated by substituting the elapsed time t after the momentary interruption up to the acquisition of the transient coil voltage value in Equation (4).

Equation (4) is an analytical solution derived using the first simplified circuit equation shown in Equation (3). Therefore, leakage current is not considered. When the leakage current is taken into consideration, an equation obtained by replacing V and the like in Equation (4) may be used as shown in Equation (5) in FIG. Also, when considering the internal resistance r of the coil 25, an equation in which R is replaced by R+2r may be used.

Further solving Equation (4) for the induced voltage V _emf yields Equation (6) in FIG. This equation can be rewritten as the equation shown in Equation (7) in FIG. That is, it becomes the general formula (1) mentioned above. However, Ka, Kb, and Kc here are those shown in addition to the formula (7).

Therefore, the momentary interruption current value and the transient coil voltage value are acquired, and together with the elapsed time t after the momentary interruption until the transient coil voltage value is acquired, the first simplified circuit equation (in detail, derived from the first simplified circuit equation), the induced voltage can be estimated.

(Selection of detection timing)
The circuit equation can be simplified by selecting the timing of momentary interruption and the timing of acquiring the transient coil voltage value.

Specifically, the momentary interruption occurs at the timing when the phase of the current reaches 90°. That is, the momentary interruption occurs at the timing when the current of the U-phase alternating current that is energized at the time of momentary interruption becomes maximum. Then, at the timing of ωt=nπ (n is a natural number), two transient coil voltage values are acquired. The difference is then obtained using the two transient coil voltage values in the circuit equation. By doing so, a simplified circuit equation (second simplified circuit equation) can be formulated.

FIG. 11 shows a first simplified circuit equation as Equation (8) in the case of an instantaneous interruption at the timing when the phase of the current reaches 90°. By setting the phase of the current to 90°, the terms of the trigonometric functions can be simplified (see formula (4) in FIG. 9).

Further, two transient coil voltage values are acquired at the timing when ωt=nπ, and the difference is obtained using the two transient coil voltage values in the formula (8). An example of the second simplified circuit equation formulated thereby is shown as Equation (9) in FIG. By doing so, the terms of the trigonometric functions can be eliminated. Therefore, the circuit equations can be simplified even further.

When adopting this method, the transient coil voltage values are acquired at two points separated by rotation intervals. Therefore, the higher the rotation speed of the motor 2, the shorter the induced voltage can be estimated. Therefore, it is effective to use this technique when the motor 2 is operating at high speed.

(Estimation of the induced voltage based on the gradient of the coil voltage at the moment of interruption)
The induced voltage can also be estimated by using the gradient (amount of change with time) of the coil voltage of the current-carrying phase at the moment of momentary interruption in the circuit equation described above. The induced voltage can be estimated in a short period of time because only instantaneous interruptions are taken into account and transients are not taken into consideration.

However, in the case of this method, it is necessary to actually measure the slope of the coil voltage at the moment of interruption. Therefore, it is effective to use it when the motor 2 is running at a low speed.

A simplified circuit equation (third simplified circuit equation) formulated by applying this technique to the first simplified circuit equation is shown in FIG. 11 as Equation (10). Since t=0, the equation can be simplified. By rearranging the formula (10) with respect to the induced voltage V _emf , the formula (11) shown in FIG. 11 is obtained. Therefore, even by using the actually measured value of the slope of the coil voltage at the time of the momentary interruption, the induced voltage can be estimated while reducing the load of the arithmetic processing and increasing the speed.

The MCU 4 selects mathematical formulas such as these, that is, basic circuit equations, simplified circuit equations, or formulas derived from these according to the specifications of the motor system 1, and stores them in the magnet temperature estimator 42 or the database 43. Remember. A plurality of formulas may be stored, or one formula may be stored.

Then, it is preferable that, among the terms constituting the formula, the part including the variable is converted into data by replacing it with a numerical value or the like by experiment or the like in advance, and the data is stored in the database 43 . Then, by using the database 43 when estimating the induced voltage, complicated arithmetic processing can be simplified and speeded up.

In particular, it is preferable to store the general formula (1) described above in the MCU4. Then, the coefficients of Ka, Kb, and Kc are converted into data by replacing them with numerical values, and the data of these coefficients are stored in the database 43 in the form of a table or the like.

Then, when estimating the induced voltage, if the current value at the moment of interruption and the coil voltage value at the time of transient are acquired, the coefficients converted into appropriate data are extracted from the database 43, and they are expressed by the general formula ( Only by substituting in 1), the induced voltage can be estimated.

<Specific example of magnet temperature estimation>
FIG. 12 illustrates a specific flow of processing for estimating the magnet temperature. The MCU 4 (inverter control unit 41 ) momentarily interrupts the operation of the motor 2 by controlling the inverter 3 . Then, the MCU 4 (magnet temperature estimating unit 42) acquires the current value at that time, that is, the current value at the time of instantaneous interruption (step S1).

Specifically, the switching elements 33U and 33L that are on-controlled are momentarily off-controlled to momentarily cut off (instantaneously cut off) the energization of the motor 2 (see FIG. 2). Since there is only a momentary interruption, the torque output by the motor 2 is hardly affected. Therefore, regardless of the operating state of the motor 2, momentary interruption can be performed as required. While the motor 2 is in operation, the power may be interrupted all the time, or at regular intervals during a predetermined period of time while the motor 2 is in operation. Since the induced voltage can be detected not only during the deceleration period of the vehicle but also during the acceleration period, it is excellent in responsiveness.

In the case of a vehicle, it is often during acceleration that the temperature of the magnet becomes particularly high. It is effective in temperature control of the magnet 23 to be able to estimate the magnet temperature at that time. Moreover, it can be estimated with high accuracy in a short time. Magnet temperature can be estimated as often as needed. Therefore, appropriate temperature control of the magnet 23 becomes easier than before. As a result, the performance of the motor 2 can be fully exhibited, so fuel and/or battery consumption can be suppressed.

At this time, if the second simplified circuit equation is used as a mathematical expression, it may be performed at the predetermined timing described above. Also, when using the third simplified circuit equation as a mathematical expression, the slope of the coil voltage may be obtained by actually measuring it at the moment of momentary interruption.

When the momentary interruption occurs, the MCU 4 acquires the coil voltage at a predetermined timing during the transition (step S2). Specifically, the MCU 4 (magnet temperature estimator 42) acquires the coil voltage of the energized phase during the momentary interruption, that is, the transient coil voltage value during the momentary interruption period. This can be done at any time during transition.

At this time, if the second simplified circuit equation is used as a formula, it may be performed twice at the above-described predetermined timing. Moreover, this processing is unnecessary when the third simplified circuit equation is used as the formula.

When the MCU 4 (magnet temperature estimating unit 42) acquires the instantaneous current value and the transient coil voltage value, the induced voltage is estimated using these values in the formula together with the time when the transient coil voltage value was acquired ( step S3). The formula to be used can be appropriately selected according to the specifications. A basic circuit equation or a simplified circuit equation may be used. Numerical formulas modified based on these may be used. When the third simplified circuit equation is used as a mathematical expression, the value of the current value at momentary interruption and the value of the slope of the coil voltage at momentary interruption may be used.

At this time, it is preferable that the portion of the formula including the variables be converted into data and stored in the database 43 as much as possible, as described above. By doing so, when estimating the induced voltage, the magnet temperature estimating unit 42 uses the database 43, so that the load of arithmetic processing can be reduced and the arithmetic processing can be speeded up.

After estimating the induced voltage, the MCU 4 (magnet temperature estimator 42) estimates the magnet temperature based on the correlation between the induced voltage and the magnet temperature (step S4). Data specifying the correlation between the induced voltage and the magnet temperature is obtained in advance by experiments or the like and stored in the database 43 . By comparing the estimated induced voltage with the data, the magnet temperature can be easily estimated.

As described above, according to the motor system 1 to which the disclosed technique is applied, the circuit equation is formulated based on the equivalent circuit model of the motor system 1 representing the behavior at the moment of transient interruption, and the induced voltage is estimated. , use the circuit equation or a formula derived from the circuit equation. As a result, the induced voltage can be detected only by actually measuring the instantaneous interruption current value and the transitional coil voltage value when the motor 2 is in operation.

Therefore, the magnet temperature can be estimated with high accuracy regardless of the operating state of the motor 2 . As a result, appropriate temperature control of the magnets 23 is facilitated, so that the performance of the motor 2 can be sufficiently exhibited, and fuel and/or battery consumption can be suppressed.

= Application example of magnet temperature estimation method =
In the magnet temperature estimating method described above, the current value at momentary interruption and the coil voltage value at transient time are actually measured, and the magnet temperature is estimated based on these values. The actual measurement requires high sensitivity and high accuracy. Therefore, various noises are superimposed on these measured values, and on the state quantities related to the estimation of the magnet temperature, such as the induced voltage and the magnet temperature obtained from these measured values. Noise affects the estimation accuracy of the magnet temperature.

A low-pass filter (LPF) may be used to process (remove) the noise, but the higher the accuracy, the lower the responsiveness. Therefore, it takes a long time to some extent to ensure accuracy in noise processing by the LPF. Therefore, when estimating the magnet temperature in an extremely short time, it is not possible to achieve both responsiveness and accuracy. Therefore, it is not preferable to use the LPF for noise processing in estimating the magnet temperature due to an instantaneous interruption.

In particular, when the motor 2 is mounted on a vehicle, there are various types of noise. The amplitude varies greatly depending on the noise. If the magnet temperature is estimated using such a state quantity superimposed with noise, there is a risk that the estimation accuracy will be lowered and the temperature of the magnet 23 will not be appropriately controlled.

In contrast, in this application example, by applying an observer, etc., even when estimating the magnet temperature for an extremely short period of time due to an instantaneous interruption, noise can be handled appropriately while maintaining both responsiveness and accuracy. , the magnet temperature estimation method is devised.

<Summary of Application>
The upper diagram of FIG. 13 shows a block diagram of a conventional observer model. u is the input, x is the state quantity, and y is the output. Each of A, B, C, and K is a matrix element forming an observer model, and K is an observer gain. The letter “^” represents an estimated value. 1/s is the integral component.

The upper side of the broken line L1 in the block diagram corresponds to the system to be controlled, and the lower side of the broken line L1 corresponds to the simulator (observer) that estimates the state quantity. The observer gain K is multiplied by the difference between the system output y and the simulator output y (estimated value), ie, the estimated error, and fed back (error feedback) to the simulator. This makes it possible to converge the state quantity estimation error. By selecting the observer gain K, the speed of convergence can be adjusted.

That is, the larger the observer gain K, the faster the convergence of the state quantity estimation error. Therefore, a large observer gain K is preferable, but if the observer gain K is large, convergence becomes unstable when noise is large. If the observer gain K is made too large, it diverges. Therefore, the observer gain K must be set to an appropriate size in consideration of convergence and stability.

Therefore, this time, as shown in the lower diagram of FIG. 13, the observer model was modified to create a new observer model (modified observer model) so that the observer gain K can be predictively changed by the input u. The magnet temperature estimation method of the application example uses this modified observer model to estimate the magnet temperature.

<Specific example of application>
FIG. 14 schematically illustrates the configuration of a magnet temperature estimation system 420 to which the magnet temperature estimation method of the applied example is applied. The portion indicated by the dashed line L2 in FIG. 14 is the main portion of the magnet temperature estimation system 420, in which the modified observer model is used. The magnet temperature estimation system 420 is composed of a first magnet temperature estimation section 421, a second magnet temperature estimation section 422, and the like. K is a feedback gain (correction coefficient) corresponding to the observer gain.

The functional components of the magnet temperature estimation system 420 may be implemented as software in the MCU 4 by extending the magnet temperature estimation section 42 described above.

A state quantity related to the estimation of the magnet 23 is input to the motor 2 . State quantities input from the inverter 3 to the motor 2 include at least current and voltage (more specifically, their phases and power are also included). Various noises are superimposed on the state quantity.

The first magnet temperature estimator 421 corresponds to the magnet temperature estimator 42 described above. The first magnet temperature estimator 421 uses such a state quantity to perform processing (first magnet temperature estimation processing) for estimating the magnet temperature based on the motor 2 (actual machine).

Specifically, similarly to the magnet temperature estimating unit 42, the first magnet temperature estimating unit 421 momentarily interrupts the energization of the coil 25 during operation of the motor 2, and the state quantity input to the motor 2 from the inverter 3 , the instantaneous power failure current value and the transient coil voltage value are obtained. Then, the induced voltage is estimated using mathematical formulas such as the circuit equations described above, the current value at the moment of interruption, and the coil voltage value at the time of transient. A magnet temperature is estimated based on the estimated induced voltage.

On the other hand, the second magnet temperature estimator 422 executes a process of estimating the magnet temperature based on the thermal model of the motor 2 (second magnet temperature estimation process). The thermal model is a dynamic heat transfer model designed to correspond to the actual motor 2 . A thermal model can be appropriately designed according to the specifications of the motor 2 .

FIG. 15 schematically shows an example of the thermal model. P represents a heat source. The heat source is a state quantity input to the motor 2 (a state quantity that generates heat, such as current or voltage). The unit of heat source is "W". Cm represents the heat capacity “J/K” of the magnet 23 . Cs represents the heat capacity “J/K” of the stator 21 . Hg and Hc each represent the heat transfer coefficient "W/K". Tc represents the temperature "K" of the motor case. Note that Hg corresponds to the gap 24 between the rotor 20 and the stator 21 .

Tm is the temperature “K” of the magnet 23 and Ts is the temperature “K” of the stator 21 . The temperature Tm of the magnet 23 and the temperature Ts of the stator 21 are input to the heat capacity of each of the magnet 23 and the stator 21 . Based on such a dynamic heat transfer model, the second magnet temperature estimator 422 estimates the magnet temperature.

As shown in FIG. 14, when a state quantity such as a current is input from the inverter 3 to the motor 2, in the first magnet temperature estimation process, based on the state quantity, induced by calculation using a circuit equation or the like. The voltage is estimated and the magnet temperature is estimated based on the induced voltage. On the other hand, in the second magnet temperature estimating process, the magnet temperature is estimated by calculating using a heat transfer model based on the state quantity.

Then, the magnet temperature estimated in the first magnet temperature estimation process is set as a first estimated value, the magnet temperature estimated in the second magnet temperature estimation process is set as a second estimated value, and these first estimated values and A difference between the second estimates, ie an estimation error, is determined. The state quantity used in the second magnet temperature estimation process is corrected with the feedback gain K according to this estimation error (error feedback). Note that the first estimated value and the second estimated value are not limited to the magnet temperature, and may be comparable estimated values.

Specifically, after the estimation error is multiplied by the feedback gain K, it is added to the state quantity used for the second magnet temperature estimation process. At this time, if the feedback gain K has an appropriate magnitude with respect to the estimation error, the estimation error will quickly converge. Both convergence and stability can be ensured. As a result, the magnet temperature can be quickly estimated with high accuracy.

(Influence of noise and measures against it)
As described above, the first magnet temperature estimating unit 421 momentarily interrupts the energization of the coil 25 while the motor 2 is running, and the current value at momentary interruption and the transient Get the current coil voltage value. Since these state quantities must be measured with high accuracy in an extremely short time, highly sensitive measurement is required.

Moreover, the magnet temperature is estimated while the vehicle is running. Therefore, various noises are superimposed on the state quantity. If the magnet temperature is estimated using such a state quantity in which various noises are superimposed, the accuracy of the first estimated value may be lowered due to the influence of the noise. As the accuracy of the first estimate decreases, the estimation error increases. As a result, the feedback gain K may become excessive with respect to the estimation error, and the error feedback may become inappropriate.

Therefore, in this application example, the magnet temperature is estimated using the modified observer model as described above. Specifically, the magnet temperature estimation system 420 predicts noise superimposed on the state quantity, and changes the magnitude of the feedback gain K according to the noise (corresponding to so-called feedforward compensation).

(noise)
Noise that poses a problem in estimating the magnet temperature can be broadly classified into noise (intrinsic noise) caused by the inverter 3 itself and noise (extrinsic noise) derived from the outside of the inverter 3 . Intrinsic noise can be predicted based on the intrinsic noise factors that cause it. Extrinsic noise can be predicted based on the extrinsic noise factors that cause it.

Intrinsic noise factors include power, current, and voltage, among others. Various electronic components are densely packed inside the inverter 3, and a large current flows. Therefore, a highly sensitive actual measurement inside the inverter 3 is likely to be affected by them. As the power input/output to/from the inverter 3 increases, so does the noise.

Motor power can also be an inherent noise factor. That is, as the output of the motor 2 increases, the power input to and output from the inverter 3 increases accordingly. Furthermore, since the motor power changes according to the accelerator opening of the vehicle, the accelerator opening can also be an inherent noise factor. The noise prediction based on the accelerator opening can be predicted at an earlier timing than the noise prediction based on the motor power.

A shift command to a transmission can also be an inherent noise factor. That is, the torque of the motor 2 may fluctuate at the timing when the gear ratio is switched. Fluctuations in torque affect power input to and output from the inverter 3, and the like. Aging deterioration of electronic components inside the inverter 3 can also be an inherent noise factor. For example, if the characteristics of the low-pass filter of the inverter 3 change, noise can be affected.

On the other hand, external noise factors include electric power of electrical equipment mounted on the vehicle. For example, the motor system 1 described above includes a power unit such as a boost converter in addition to the inverter 3 . Power to and from these can also affect noise.

The power of the auxiliary system (so-called 12V voltage system with lead-acid batteries) can also be an external noise factor. Since the accessory system is also electrically connected to the motor system 1, large fluctuations in its power can affect noise. For example, when the power windows are activated, a relatively large amount of power is consumed, which is likely to affect noise.

Furthermore, vehicles travel in a variety of environments. Therefore, noise (specific environment noise) may occur due to a specific environment. For example, a vehicle may run in an environment with a strong electric field, such as near a radio tower. In such cases, noise may be affected. Therefore, the specific environment can also be an external noise factor.

A noise factor to be predicted is appropriately selected from among such intrinsic noise factors and extrinsic noise factors, and preset in the magnet temperature estimation system 420 . The magnet temperature estimation system 420 predicts noise superimposed on the state quantity based on the set noise factor. Then, magnet temperature estimation system 420 changes feedback gain K according to the noise.

For example, power, current, and voltage are state quantities input from the inverter 3 to the motor 2, and are direct intrinsic noise factors. As the power or the like increases, the noise superimposed thereon increases accordingly. If the power or the like becomes smaller, the noise superimposed thereon becomes smaller accordingly. That is, there is a positive correlation between the magnitude of power and the magnitude of noise superimposed thereon.

Therefore, in the magnet temperature estimation system 420, as shown in FIG. 14, the feedback gain K is changed when electric power or the like is input from the inverter 3 to the motor 2. FIG. Specifically, the feedback gain K is changed to decrease as the current and/or voltage increase.

FIG. 16 shows the correspondence between feedback gain K and current and voltage. K1 is a feedback gain (first feedback gain) corresponding to current, and K2 is a feedback gain (second feedback gain) corresponding to voltage. The feedback gain K in this application is the product of the first feedback gain K1 and the second feedback gain K2.

The first feedback gain K1 is changed so as to decrease as the current increases. The second feedback gain K2 is also changed so as to decrease as the voltage increases. The ratio of the change in the first feedback gain K1 to the change in current, that is, the slope of the graph G1 is appropriately set according to the specifications of the motor system 1. FIG. Similarly, the ratio of the change in the second feedback gain K2 to the change in voltage, that is, the slope of the graph G2 is also appropriately set according to the specifications of the motor system 1. FIG. These graphs G1 and G2 are not limited to straight lines, and may be curved lines.

Also, the slopes of these graphs G1 and G2 may be different or the same. However, current is more likely to be superimposed with large noise than voltage, so it is preferable that the slope of the graph is larger for current than for voltage. Data such as a table specifying such a correspondence relationship is stored in the database 43 of the MCU 4 and used as necessary when estimating the magnet temperature.

As indicated by the solid line L3 in FIG. 14, the feedback gain K is predictively changed from the state quantity input to the magnet temperature estimation system 420. Therefore, even if noise is superimposed on the state quantity, In addition, the feedback gain K can be changed according to its magnitude. As a result, error feedback can be appropriately performed. Even if a large amount of noise is superimposed on the state quantity, both convergence and stability can be achieved in an appropriate state. As a result, the magnet temperature can be quickly estimated with high accuracy.

Even when the noise factor to be predicted is other than power, etc., the countermeasures against noise are the same as in the case of power. That is, data specifying the correspondence relationship between the noise factor and the noise magnitude is obtained in advance through experiments or the like, and stored in the database 43 of the MCU 4 .

Then, when estimating the magnet temperature, the magnet temperature estimation system 420 predicts the noise superimposed on the state quantity based on the data, and as shown by the two-dot chain line L4 in FIG. The magnitude of the gain K may be changed.

<Specific example of estimating the magnet temperature in the application example>
FIG. 17 illustrates the flow of processing for estimating the magnet temperature in the application. When the operation of the vehicle is started and the output of the motor 2 is requested, the MCU 4 (inverter control unit 41) starts the operation of the motor 2 by controlling the inverter 3 (step S10).

Then, when magnet temperature estimation is requested, the MCU 4 (magnet temperature estimation system 420) estimates the magnet temperature (step S11). As described above, the magnet temperature can be estimated regardless of the operating state of the motor 2 .

When estimating the magnet temperature, the MCU 4 executes a first magnet temperature estimation process and a second magnet temperature estimation process (steps S12 and S13). The MCU 4 also predicts noise superimposed on the electric power, etc., as described above, and determines whether or not the feedback gain K needs to be changed (step S14). It is preferable to perform each process of these steps S12, S13, and S14 simultaneously in parallel.

Then, when determining that the feedback gain K needs to be changed, the MCU 4 changes the magnitude of the feedback gain K according to the magnitude of the noise (step S15). At this time, if the magnitude of the feedback gain K is to be changed according to the magnitude of the electric power or the like, it may be changed according to the magnitude of the electric power or the like input to the motor 2 as described above. In this case, determination of necessity of change can be omitted. The magnitude of the feedback gain K should be changed according to the predicted noise. On the other hand, when the MCU 4 determines that the feedback gain K does not need to be changed, the MCU 4 directly proceeds to the next step S17.

The MCU 4 also acquires an estimation error by executing the first magnet temperature estimation process and the second magnet temperature estimation process (step S16). Then, the MCU 4 performs error feedback using the obtained estimation error and feedback gain K, and corrects the state quantity used in the second magnet temperature estimation process (step S17).

By doing so, the estimation error converges stably and quickly, and the magnet temperature is quickly and accurately estimated (step S18). Since the feedback gain K is changed to an appropriate magnitude according to the magnitude of the noise, the magnet temperature can be quickly and accurately estimated even if a large amount of noise is superimposed on the power or the like.

The MCU 4 also compares the estimated magnet temperature with a predetermined threshold Tm (step S19). The threshold Tm is a value close to the upper limit of the magnet temperature, and is stored in the database 43 of the MCU4. That is, since the magnet temperature can be estimated quickly and accurately, it is possible to control the temperature at a value close to the upper limit of the magnet temperature.

As a result, when determining that the estimated magnet temperature is higher than the threshold value Tm, the MCU 4 limits the current amount of the inverter 3 (step S20). By doing so, an increase in the magnet temperature is suppressed, and it is possible to prevent the magnet temperature from reaching the upper limit.

As described above, according to the motor system 1 of the application example to which the disclosed technology is applied, in addition to the magnet temperature estimation method due to the momentary interruption described above, the noise superimposed on the electric power or the like is predictively estimated and appropriately processed. Therefore, it is possible to suppress the influence of noise on estimation of the magnet temperature in an extremely short time.

Therefore, regardless of the operating state of the motor 2, the magnet temperature can be estimated with higher accuracy. As a result, appropriate temperature control of the magnets 23 is facilitated, so that the performance of the motor 2 can be sufficiently exhibited, and fuel and/or battery consumption can be suppressed.

<Demonstration test>
A test was conducted to verify the effect of the disclosed technology. In the test, a predetermined test machine was used, and a telemeter was attached to the motor so that the magnet temperature could be measured.

In the test, the motor was rotated at 1000 rpm, and in this state, an instantaneous interruption of several milliseconds was repeatedly performed at intervals of milliseconds. Then, the magnet temperature was estimated using the above-described magnet temperature estimation system 420 for each momentary interruption. Since it is a test machine, there is no problem of large noise. Therefore, the feedback gain K remains unchanged. A summary of the test results is shown in FIG.

The graph of the measured values shown in FIG. 18 represents the change over time of the measured values of the magnet temperature. A graph of estimated values represents changes over time in the estimated values of the magnet temperature estimated by the magnet temperature estimation system 420 . These are changes over time in seconds.

As these graphs show, there was a tendency for the estimated values to gradually approach the measured values. The maximum error in this demonstration test was about 5°C. Therefore, by using the disclosed technology, it has been demonstrated that the magnet temperature can be estimated with high accuracy regardless of the operating state of the motor.

Note that the technology disclosed is not limited to the above-described embodiments, and includes various other configurations.

For example, in the above-described embodiments, an in-vehicle motor system was exemplified, but the motor system to which the technology disclosed can be applied is not limited to in-vehicle systems.

In the above-described embodiments, a plurality of simplified circuit equations obtained by applying a predetermined simplification technique to the basic circuit equations are shown, but these are merely examples. Simplified circuit equations other than those illustrated can be formulated by appropriately combining the basic circuit equations with the simplification methods described above. A simplified circuit equation may be appropriately selected according to the specifications of the motor system. Also, the exemplified formula itself may not be used. The exemplified formula may be modified as appropriate and used.

The intrinsic and extrinsic noise factors listed in the application example are an example. Anything that causes noise superimposed on the state quantity associated with magnet temperature estimation can be a noise factor.

1 motor system 2 motor 3 inverter 4 motor control unit (MCU)
5 DC power supply 20 Rotor 21 Stator 22 Shaft 23 Magnet 24 Gap 25 Coil 26 Connection cable 30 Power supply cable 31 Inverter circuits 33U, 33L Switching element 34 Diode 41 Inverter control unit 42 Magnet temperature estimation unit 43 Database 420 Magnet temperature estimation system 421 First Magnet temperature estimator 422 Second magnet temperature estimator

Claims (6)

A method for estimating a magnet temperature of a motor in which magnets forming magnetic poles are installed in a rotor and rotated by supplying AC current formed by controlling an inverter to a plurality of coils of a stator, the method comprising:
a first magnet temperature estimation process for estimating the temperature of the magnet based on the actual motor using the state quantity input from the inverter to the motor; and a second magnet temperature estimation process for estimating the temperature, and
According to the difference between the first estimated value estimated by the first magnet temperature estimation process and the second estimated value estimated by the second magnet temperature estimation process, in the second magnet temperature estimation process performing error feedback for correcting the state quantity to be used with a predetermined correction coefficient;
A magnet temperature estimating method, wherein noise superimposed on the state quantity is predicted, and the correction coefficient is changed according to the noise. In the magnet temperature estimation method according to claim 1,
The state quantity is composed of at least one of current and voltage,
A method for estimating magnet temperature, wherein the correction coefficient is changed so as to decrease as the state quantity increases. In the magnet temperature estimation method according to claim 1 or 2,
As the first temperature estimation process,
momentarily interrupting the energization of the coil during operation of the motor;
Obtaining a current value at momentary interruption, which is the current value at the moment of momentary interruption, and a coil voltage value at transient, which is the voltage value of the coil at the moment of momentary interruption,
estimating an induced voltage using a predetermined formula formulated based on the voltage change of the coil after the momentary interruption, the current value at the time of momentary interruption and the coil voltage value at the time of transient,
A magnet temperature estimation method for estimating the temperature of the magnet based on the calculated induced voltage. In the magnet temperature estimation method according to any one of claims 1 to 3,
The magnet temperature estimation method, wherein the noise includes inherent noise caused by the inverter itself, and the magnitude of the correction coefficient is changed according to the inherent noise. In the magnet temperature estimation method according to any one of claims 1 to 4,
The motor is mounted as a drive source on an electric vehicle,
A magnet temperature estimating method including external noise caused by an electrical component other than the inverter mounted on the electric vehicle, and changing the magnitude of the correction coefficient according to the external noise. In the magnet temperature estimation method according to any one of claims 1 to 5,
The motor is mounted as a drive source on an electric vehicle,
A magnet temperature estimation method including specific environmental noise generated due to a specific environment in which the electric vehicle travels, and changing the magnitude of the correction coefficient according to the specific environmental noise.

Images