A speed loop control method for smooth magnetic field regulation of adjustable flux permanent magnet synchronous motor

Abstract

A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor belongs to the field of motor control. The present invention solves the problem of drastic changes in reluctance torque during online magnetic regulation of an adjustable flux permanent magnet synchronous motor. The present invention includes the following steps: 1. Using the inductance parameters and permanent magnet flux linkage that accompany the changes in magnetization state and direct-axis current as a data set to train a neural network model; 2. Obtaining the motor's quadrature and direct-axis currents based on the collected stator current and rotor position information, and obtaining predicted values ​​of the motor's inductance parameters and permanent magnet flux linkage through the trained model; 3. Inputting the motor's quadrature and direct-axis currents, the predicted inductance parameter values, and the predicted permanent magnet flux linkage values ​​into a torque-current gain adjuster to obtain predicted values ​​of corresponding components of the reluctance torque and an adjusted torque-current gain; 4. Obtaining a given quadrature-axis current value to further obtain a drive signal for an inverter; and repeating steps 2 to 4 to suppress torque-speed pulsation during magnetic regulation.

Description

Technical Field

[0001] The invention relates to a speed loop control method for smooth magnetic regulation of an adjustable magnetic flux permanent magnet synchronous motor, belonging to the field of motor control. Background Art

[0002] Due to their advantages such as high efficiency, high power density, and high power factor, permanent magnet synchronous motors have become the mainstream motor type for electric vehicles. However, some problems also exist. For example, the permanent magnet flux linkage of traditional permanent magnet synchronous motors is difficult to adjust, resulting in a narrow constant power range and a limited speed operating range. (If flux-weakening control is used to increase speed, the continuously applied direct-axis flux-weakening current will cause significant losses in the system, resulting in low efficiency.) Adjustable-flux permanent magnet synchronous motors offer additional degrees of freedom in adjusting the permanent magnet flux linkage. Since their introduction, these motors have been widely studied and received significant attention in the field of wide-speed motor regulation. In recent years, research by domestic and foreign scholars on the topology, electromagnetic properties, and magnetic control of adjustable-flux permanent magnet synchronous motors has shown that adjustable-flux permanent magnet synchronous motors have advantages such as a wide speed regulation range, high high-speed operating efficiency, a large proportion of the high-efficiency zone, and the ability to reduce the use of rare earth permanent magnet materials. These motors have good application prospects in the field of wide speed regulation.

[0003] Currently, the magnetization process for adjustable-flux permanent magnet synchronous motors (CFMs) is mostly static. This involves applying a magnetization pulse current to adjust the magnetization state of the permanent magnets while the motor is static (rotor locked or stationary). To enable CFMs for electric vehicle applications, ensuring stable online magnetization is a critical challenge. However, CFMs require large direct-axis pulse currents to adjust the rotor's permanent magnet flux. This causes the reluctance torque of the CFM to fluctuate dramatically during online magnetization, with its magnitude even exceeding the permanent magnet torque. When the reluctance torque and permanent magnet torque are in opposite directions, increasing the quadrature-axis current causes the motor to generate a reverse torque. In this case, applying a traditional dual-loop speed and current control strategy will cause system instability, making it impossible to maintain a stable motor speed. This results in large torque and speed fluctuations during the motor's magnetization process, making stable online magnetization difficult. Therefore, improved control strategies are necessary to cope with the dramatic fluctuations in reluctance torque during online magnetization. Summary of the Invention

[0004] In order to solve the problem of drastic changes in reluctance torque during online magnetization of an adjustable flux permanent magnet synchronous motor, the present invention provides a speed loop control method for smooth magnetization of an adjustable flux permanent magnet synchronous motor.

[0005] The present invention provides a speed loop control method for smooth magnetic field regulation of an adjustable flux permanent magnet synchronous motor, the method comprising the following steps:

[0006] Step 1: Build a finite element model of an adjustable flux permanent magnet synchronous motor. Simulate the inductance parameters and permanent magnet flux linkage of the adjustable flux permanent magnet synchronous motor as the magnetization state and direct-axis current change. Use this as a data set to train a neural network model and deploy it on the motor controller.

[0007] Step 2: Collect the stator current and rotor position information of the motor as sampling signals, perform coordinate transformation on the sampling signals to obtain feedback of the motor's direct-axis current and quadrature-axis current, and further obtain predicted values ​​of the motor's inductance parameters and permanent magnet flux through the neural network model trained in step 1;

[0008] Step 3: Input the motor direct-axis current, quadrature-axis current, predicted values ​​of the motor inductance parameters, and predicted values ​​of the permanent magnet flux linkage into the torque-current gain adjuster to obtain the predicted values ​​of the corresponding components of the reluctance torque and the adjusted torque-current gain;

[0009] Step 4: Input the predicted value of the corresponding component of the reluctance torque, the actual motor speed, and the set speed into the linear active disturbance rejection speed controller. The output of the linear active disturbance rejection speed controller is multiplied by the adjusted torque-current gain as the quadrature-axis current set value of the current loop input. The current loop output is subjected to voltage-current decoupling and, after coordinate transformation and pulse width modulation, a drive signal for the inverter is obtained.

[0010] Repeat steps 2 to 4 so that the actual motor speed tracks the reference value to suppress torque and speed pulsation during magnetic modulation.

[0011] Preferably, the neural network model in step 1 adopts a BP neural network model, and the model input is the direct axis current and the predicted value of permanent magnet flux in the previous control cycle , the model output is the predicted value of direct-axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle , the predicted value of permanent magnet flux linkage in the previous control cycle The initial value is the permanent magnet flux linkage in the initial magnetization state when the motor initially operates in the highest magnetization state.

[0012] Preferably, the specific process of step 2 is:

[0013] Collect the stator current of the motor 、 、 With rotor position information As the sampling signal, the sampling signal is subjected to Clark transformation and Park transformation to obtain the motor direct axis current , quadrature-axis current ; The direct axis current and the predicted value of permanent magnet flux in the previous control cycle Input into the neural network model trained in step 1 to obtain the predicted value of direct axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle .

[0014] Preferably, the specific process of step three is:

[0015] Predicted direct-axis inductance , predicted value of quadrature-axis inductance , direct axis current , quadrature-axis current And the predicted value of permanent magnet flux in the current control cycle Common input to the torque-current gain adjuster;

[0016] Among them, the predicted value of the corresponding component of the reluctance torque is Get it as follows:

[0017]

[0018] Where, is the motor moment of inertia, is the number of motor pole pairs;

[0019] Among them, the adjusted torque-current gain gain The acquisition process is:

[0020] calculate , limit it to [-1,1], and finally discretize the value to get the torque-current gain gain ;

[0021] Where, It is the permanent magnet flux value of the motor in the highest magnetization state.

[0022] Preferably, the quadrature axis current given value of the current loop input is The acquisition is realized by using a linear active disturbance rejection speed controller, which includes an extended state observer, a nonlinear feedback and an adder, and the quadrature axis current given value The acquisition process is:

[0023] The predicted value of the corresponding component of the reluctance torque , Actual motor speed speed and the product bu Input to the extended state observer, where b is the compensation coefficient, u is the output of the linear active disturbance rejection speed controller, and the extended state observer outputs the disturbance prediction value f_pre And the motor speed prediction value speed_p ; The motor speed prediction value speed_p With the motor speed given value speed_ref Difference, perform nonlinear operations on it, and then subtract get u , further u With the adjusted torque-current gain gain Multiply to get the given value of the quadrature axis current .

[0024] Preferably, the compensation coefficient b The value of is updated as follows:

[0025] .

[0026] Preferably, the input of the current loop in step 4 includes a given value of the quadrature axis current and the direct-axis current given value , during the magnetic modulation period, the direct axis current is given A trapezoidal pulse is generated according to the target magnetization state, and the direct axis current is set to a given value at other times. Set to 0;

[0027] Will 、 As the input of the PI current controller, the voltage and current of the PI current controller output are decoupled, and after Park inverse transformation and space vector pulse width modulation, the driving signal is obtained and input into the inverter to drive the motor.

[0028] The present invention takes into account the dramatic changes in reluctance torque during the field modulation of an adjustable-flux permanent magnet synchronous motor. It utilizes a neural network to predict the motor's inductance parameters and permanent magnet flux linkage in real time, further calculating the predicted reluctance torque. This prediction is then incorporated into a linear state expansion observer (SEO), enabling the SEO to accurately obtain observed values ​​during field modulation while ensuring high dynamic response and effectively suppressing torque and speed fluctuations during field modulation.

[0029] A simulation model was built to compare the proposed speed loop control method for smooth magnetization of adjustable flux permanent magnet synchronous motor with the traditional PI speed control strategy. Demagnetization pulse current was applied at 0.3~0.32s, and magnetization pulse current was applied at 0.95~1s. Other conditions such as speed set value and load torque were the same.

[0030] Applying the traditional control strategy, the speed waveform is as follows: Figure 4 As shown, it can be seen that during the magnetic adjustment period (especially the magnetization period), the adjustable flux permanent magnet synchronous motor will produce large speed fluctuations.

[0031] The speed waveform of the present invention is as follows: Figure 5As shown, it can be seen that the proposed control strategy not only has excellent dynamic performance, but also can effectively suppress torque and speed fluctuations during magnetic modulation to ensure smooth operation of the motor. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is the schematic diagram of the torque-current gain regulator;

[0033] Figure 2 This is the schematic diagram of the BP neural network model;

[0034] Figure 3 It is a schematic diagram of a speed controller involved in the method of the present invention;

[0035] Figure 4 The speed waveform is obtained by applying the traditional control strategy;

[0036] Figure 5 This is a rotation speed wave diagram obtained by applying the method of the present invention;

[0037] Figure 6 It is a control block diagram of a specific embodiment. DETAILED DESCRIPTION

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0039] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0040] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0041] Specific implementation method 1: Figures 1 to 6 This embodiment describes a speed loop control method for smooth magnetic field regulation of an adjustable flux permanent magnet synchronous motor, the method comprising the following steps:

[0042] Step 1: Build a finite element model of an adjustable flux permanent magnet synchronous motor. Simulate the inductance parameters and permanent magnet flux linkage of the adjustable flux permanent magnet synchronous motor as the magnetization state and direct-axis current change. Use this as a data set to train a neural network model and deploy it on the motor controller.

[0043] Step 2: Collect the stator current and rotor position information of the motor as sampling signals, perform coordinate transformation on the sampling signals to obtain feedback of the motor's direct-axis current and quadrature-axis current, and further obtain predicted values ​​of the motor's inductance parameters and permanent magnet flux through the neural network model trained in step 1;

[0044] Step 3: Input the motor direct-axis current, quadrature-axis current, predicted values ​​of the motor inductance parameters, and predicted values ​​of the permanent magnet flux linkage into the torque-current gain adjuster to obtain the predicted values ​​of the corresponding components of the reluctance torque and the adjusted torque-current gain;

[0045] Step 4: Input the predicted value of the corresponding component of the reluctance torque, the actual motor speed, and the set speed into the linear active disturbance rejection speed controller. The output of the linear active disturbance rejection speed controller is multiplied by the adjusted torque-current gain as the quadrature-axis current set value of the current loop input. The current loop output is subjected to voltage-current decoupling and, after coordinate transformation and pulse width modulation, a drive signal for the inverter is obtained.

[0046] Repeat steps 2 to 4 so that the actual motor speed tracks the reference value to suppress torque and speed pulsation during magnetic modulation.

[0047] First, the design concept of the present invention is explained. The method of the present invention uses a speed controller to regulate the speed of a permanent magnet synchronous motor. The speed controller is designed based on the linear active disturbance rejection control theory. A linear state expansion observer is designed to observe disturbances and motor speed. Nonlinear feedback is used to optimize the dynamic performance of speed control. Controller parameters such as the compensation coefficient and the predicted reluctance torque are based on the inductance parameters and the dynamic changes of the permanent magnet flux predicted by the neural network. The speed controller design process is as follows:

[0048] The mechanical equation and electromagnetic torque equation of the adjustable flux permanent magnet synchronous motor are as follows

[0049]

[0050] Where, is the motor moment of inertia, B is the damping coefficient of the motor, is the electromagnetic torque of the motor, is the motor load torque, is the motor rotation angular velocity, is the predicted value of permanent magnet flux in the current control period, is the number of motor pole pairs, 、 is the motor direct-axis and quadrature-axis inductance, 、 are the direct-axis and quadrature-axis currents of the motor.

[0051] Combined

[0052]

[0053] Among them, the corresponding component of the reluctance torque is

[0054]

[0055] So set the state variable 、 As follows, 、 are the actual value and predicted value of the corresponding component of the reluctance torque, Represents the total disturbance value of speed control.

[0056]

[0057] Based on this, the LADR speed controller is designed as follows, where b is the compensation coefficient and u is the speed controller output. 、 、 Both correspond 、 、 The derivative of .

[0058]

[0059] Easy to get compensation coefficient Predicted value of the component corresponding to the reluctance torque As follows, 、 、 They are the direct-axis inductance prediction value, the quadrature-axis inductance prediction value, and the permanent magnet flux linkage prediction value of the current control cycle, all of which are output by the neural network model.

[0060]

[0061] The corresponding linear extended state observer is designed as follows, where 、 The observer is the state variable 、 The observed value of Corresponding observer output (here set with is equal to the observed value of the motor speed). 、 is the internal parameter of the observer (which can be set according to the bandwidth method).

[0062]

[0063] Then, based on the above design concept, the method of the present invention is described in detail.

[0064] See also Figure 2 The neural network model in step 1 adopts BP neural network model, and the model input is direct axis current and the predicted value of permanent magnet flux in the previous control cycle , the model output is the predicted value of direct-axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle , the predicted value of permanent magnet flux linkage in the previous control cycle The initial value is the permanent magnet flux linkage in the initial magnetization state when the motor initially operates in the highest magnetization state.

[0065] Predicted direct-axis inductance , predicted value of quadrature-axis inductance are the predicted values ​​of the inductor parameters.

[0066] The specific process of step 2 is:

[0067] Collect the stator current of the motor 、 、 With rotor position information As the sampling signal, the sampling signal is subjected to Clark transformation and Park transformation to obtain the motor direct axis current , quadrature-axis current ; The direct axis current and the predicted value of permanent magnet flux in the previous control cycle Input into the neural network model trained in step 1 to obtain the predicted value of direct axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle .

[0068] See also Figure 1 , the specific process of step three is:

[0069] Predicted direct-axis inductance , predicted value of quadrature-axis inductance , direct axis current , quadrature-axis current And the predicted value of permanent magnet flux in the current control cycle Common input to the torque-current gain adjuster;

[0070] Among them, the predicted value of the corresponding component of the reluctance torque is Get it as follows:

[0071]

[0072] Where, is the motor moment of inertia, is the number of motor pole pairs;

[0073] Among them, the adjusted torque-current gain gain The acquisition process is:

[0074] calculate , limit it to [-1,1], and finally discretize the value and output it gain ;

[0075] Where, It is the permanent magnet flux value of the motor in the highest magnetization state.

[0076] See also Figure 3 , the quadrature axis current given value of the current loop input in step 4 The acquisition is realized by using a linear active disturbance rejection speed controller, which includes an extended state observer, a nonlinear feedback and an adder, and the quadrature axis current given value The acquisition process is:

[0077] The predicted value of the corresponding component of the reluctance torque , Actual motor speed speed and the product bu Input to the extended state observer, where b is the compensation coefficient, u is the output of the linear active disturbance rejection speed controller, and the extended state observer outputs the disturbance prediction value f_pre And the motor speed prediction value speed_p ; The motor speed prediction value speed_p With the motor speed given value speed_ref Difference, perform nonlinear operations on it, and then subtract get u , further u With the adjusted torque-current gain gain Multiply to get the given value of the quadrature axis current .

[0078] Compensation coefficient b The value of is updated as follows:

[0079] .

[0080] The input of the current loop in step 4 includes the quadrature axis current given value and the direct-axis current given value , during the magnetic modulation period, the direct axis current is given A trapezoidal pulse is generated according to the target magnetization state, and the direct axis current is set to a given value at other times. Set to 0;

[0081] Will 、 As the input of the PI current controller, the voltage and current of the PI current controller output are decoupled, and after Park inverse transformation and space vector pulse width modulation, the driving signal is obtained and input into the inverter to drive the motor.

[0082] In each control cycle of the method of the present invention, the working process of the speed controller is as follows: the direct axis current is obtained by performing coordinate transformation on the motor stator current sampling signal. , quadrature-axis current The motor initially works in the highest magnetization state, and the initial magnetization state is compared with the direct axis current. Input the BP neural network model to obtain the predicted value of direct axis inductance , predicted value of quadrature-axis inductance , the predicted value of permanent magnet flux in the current control cycle , and each time after that the predicted value of permanent magnet flux linkage in the previous control cycle As the input of the model, 、 、 After the motor permanent magnet flux and inductance parameters are updated, use Update compensation coefficient b , and 、 、 、 、 Input the torque-current gain regulator to obtain the gain of the torque-current gain regulator output gain And the predicted value of the corresponding component of the reluctance torque ; The predicted value of the corresponding component of the reluctance torque , Actual motor speed speed and compensation coefficient b With the ADRC output u The product of is input into the linear extended state observer, and the linear extended state observer outputs the disturbance prediction value and the motor speed prediction value speed_p ; The motor speed prediction value speed_p With the motor speed given value speed_ref The difference is made, and nonlinear operation is performed to improve the dynamic performance. Then the disturbance prediction value output by the linear extended state observer and the prediction value of the corresponding component of the reluctance torque are subtracted to obtain u , further combined with the gain of the torque-current gain regulator output gain Multiplying them gives the given value of the quadrature-axis current.

[0083] The following combination Figure 6 Two examples are given.

[0084] Example 1:

[0085] The present invention is applicable to the drive control scenario of online magnetic field and speed regulation of adjustable flux permanent magnet synchronous motors, and proposes a speed loop control method for smooth magnetic field regulation of adjustable flux permanent magnet synchronous motors. Figure 6 The motor control block diagram using this control method is shown. The specific steps are as follows:

[0086] Step 1: Build a corresponding adjustable flux permanent magnet synchronous motor model. Finite element simulation is used to obtain the inductance parameters and permanent magnet flux linkage of the adjustable flux permanent magnet synchronous motor under different currents and initial magnetization states. This data set is used to train a BP neural network model and deploy it on a microcontroller.

[0087] Step 2: Collect the stator current of the motor 、 、 With rotor position information As the sampling signal, the sampling signal is subjected to Clark transformation to obtain αβ Coordinate system (two-phase stationary coordinate system) i α and i β Then, Park transformation is performed to obtain the feedback of the motor direct-axis and quadrature-axis currents. and , and further obtain the predicted motor inductance parameters through the neural network model 、 and the predicted value of permanent magnet flux in the current control cycle ;

[0088] Step 3: 、 、 、 and Input torque-current gain adjuster to obtain the predicted value of reluctance torque With the adjusted torque-current gain gain ;

[0089] Step 4: Predict the reluctance torque , Actual motor speed speed With given speed speed_ref Input to the linear ADRC speed controller, the output of the linear ADRC speed controller is multiplied by the adjusted torque-current gain gain As the quadrature-axis current setpoint , during the magnetic modulation period, the direct axis current is given A trapezoidal pulse is generated according to the target magnetization state, and the direct axis current is set to a given value at other times. Set to 0;

[0090] Step 5: 、 As the input of the PI current controller, the feedforward voltage is superimposed on the PI controller output U d_ff and U q_ff Decoupling voltage and current is performed to obtain dq Reference voltage in the coordinate system (two-phase rotating coordinate system) and , and after Park inverse transformation, we get αβ Reference voltage in the coordinate system U α and U β Space vector pulse width modulation is used to obtain a driving signal which is input into the inverter to drive the motor;

[0091] Step 6: Repeat steps 2 to 5 so that the actual speed value tracks the reference value, so that the motor can run stably throughout the entire operation process (including the magnetization state adjustment period).

[0092] Example 2:

[0093] The motor control block diagram used in this embodiment is the same as that in the first embodiment, and the specific steps are as follows:

[0094] Step 1: Establish a corresponding adjustable flux permanent magnet synchronous motor model. Through finite element simulation, obtain the inductance parameters and permanent magnet flux linkage of the adjustable flux permanent magnet synchronous motor under different currents and initial magnetization states. This is used as a sample data set to train, test, and verify the neural network model. Use the BP algorithm to adjust the weight and bias of each neuron in the reverse direction according to the gradient of the error between the neural network output value and the sample value. When the error is less than the set value, the training is completed. The neural network structure and the weight and bias parameters of each neuron are determined, and then deployed on the microcontroller accordingly.

[0095] Step 2: Collect the stator current of the motor 、 、 With rotor position information As the sampling signal, the sampling signal is subjected to Clark transformation to obtain αβ In the coordinate system i α and i β Then, Park transformation is performed to obtain the feedback of the motor direct-axis and quadrature-axis currents. and , and further obtain the predicted motor inductance parameters through the neural network model 、 and the predicted value of permanent magnet flux in the current control cycle ;

[0096] Step 3: 、 、 、 and Input torque-current gain adjuster to obtain the predicted value of reluctance torque With the adjusted torque-current gain gain ;

[0097] Step 4: Predict the reluctance torque , Actual motor speed speed With given speed speed_ref Input to the linear ADRC speed controller, the output of the linear ADRC speed controller is multiplied by the adjusted torque-current gain gain As the quadrature-axis current setpoint , during the magnetic modulation period, the direct axis current is given A trapezoidal pulse is generated according to the target magnetization state, and at other times according to the predicted motor inductance parameters and , the predicted value of permanent magnet flux in the current control cycle , given value of quadrature-axis current To update the direct axis current given value To achieve MTPA control;

[0098] Step 5: 、 As the input of the PI current controller, the feedforward voltage is superimposed on the PI controller output U d_ff and U q_ff Decoupling voltage and current is performed to obtain dq Reference voltage in the coordinate system and , and after Park inverse transformation, we get αβ Reference voltage in the coordinate system U α and U β Space vector pulse width modulation is used to obtain a driving signal which is input into the inverter to drive the motor;

[0099] Step 6: Repeat steps 2 to 5 so that the actual speed value tracks the reference value, so that the motor can run stably throughout the entire operation process (including the magnetization state adjustment period).

[0100] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely illustrative of the principles and applications of the invention. It should be understood that many modifications may be made to the illustrative embodiments, and that other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in ways other than those described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in conjunction with other described embodiments.

Claims

1. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor, characterized in that: The method comprises the following steps: Step 1: Build a finite element model of an adjustable flux permanent magnet synchronous motor. Simulate the inductance parameters and permanent magnet flux linkage of the adjustable flux permanent magnet synchronous motor as the magnetization state and direct-axis current change. Use this as a data set to train a neural network model and deploy it on the motor controller. Step 2: Collect the stator current and rotor position information of the motor as sampling signals, perform coordinate transformation on the sampling signals to obtain feedback of the motor's direct-axis current and quadrature-axis current, and further obtain predicted values ​​of the motor's inductance parameters and permanent magnet flux through the neural network model trained in step 1; Step 3: Input the motor direct-axis current, quadrature-axis current, predicted values ​​of the motor inductance parameters, and predicted values ​​of the permanent magnet flux linkage into the torque-current gain adjuster to obtain the predicted values ​​of the corresponding components of the reluctance torque and the adjusted torque-current gain; Step 4: Input the predicted value of the corresponding component of the reluctance torque, the actual motor speed, and the set speed into the linear active disturbance rejection speed controller. The output of the linear active disturbance rejection speed controller is multiplied by the adjusted torque-current gain as the quadrature-axis current set value of the current loop input. The current loop output is subjected to voltage-current decoupling and, after coordinate transformation and pulse width modulation, a drive signal for the inverter is obtained. Repeat steps 2 to 4 so that the actual motor speed tracks the reference value to suppress torque and speed pulsation during magnetic modulation.

2. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 1, characterized in that: The neural network model in step 1 adopts BP neural network model, and the model input is direct axis current and the predicted value of permanent magnet flux in the previous control cycle , the model output is the predicted value of direct-axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle , the predicted value of permanent magnet flux linkage in the previous control cycle The initial value is the permanent magnet flux linkage in the initial magnetization state when the motor initially operates in the highest magnetization state.

3. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 2, characterized in that: The specific process of step 2 is: Collect the stator current of the motor 、 、 With rotor position information As the sampling signal, the sampling signal is subjected to Clark transformation and Park transformation to obtain the motor direct axis current , quadrature-axis current ; The direct axis current and the predicted value of permanent magnet flux in the previous control cycle Input into the neural network model trained in step 1 to obtain the predicted value of direct axis inductance , predicted value of quadrature-axis inductance And the predicted value of permanent magnet flux in the current control cycle .

4. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 3, characterized in that: The specific process of step three is: Predicted direct-axis inductance , predicted value of quadrature-axis inductance , direct axis current , quadrature-axis current And the predicted value of permanent magnet flux in the current control cycle Common input to the torque-current gain adjuster; Among them, the predicted value of the corresponding component of the reluctance torque is Get it as follows: Where, is the motor moment of inertia, is the number of motor pole pairs; Among them, the process of obtaining the adjusted torque-current gain is: calculate , limit it to [-1,1], and finally discretize the value to obtain the torque-current gain; Where, It is the permanent magnet flux value of the motor in the highest magnetization state.

5. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 4, characterized in that: The quadrature-axis current reference value as the current loop input The acquisition is realized by using a linear active disturbance rejection speed controller, which includes an extended state observer, a nonlinear feedback and an adder, and the quadrature axis current given value The acquisition process is: The predicted value of the corresponding component of the reluctance torque , the actual motor speed speed and the product bu are input to the extended state observer, where b is the compensation coefficient, u is the output of the linear active disturbance rejection speed controller, and the extended state observer outputs the disturbance prediction value f_pre and the motor speed prediction value speed_p; the motor speed prediction value speed_p is subtracted from the motor speed reference value speed_ref, and a nonlinear operation is performed on it, and then subtracted Get u, and further multiply u with the adjusted torque-current gain to get the quadrature-axis current given value .

6. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 5, characterized in that: The value of the compensation coefficient b is updated as follows: 。 7. A speed loop control method for smooth magnetic regulation of an adjustable flux permanent magnet synchronous motor according to claim 5, characterized in that: The input of the current loop in step 4 includes the quadrature axis current given value and the direct-axis current given value , during the magnetic modulation period, the direct axis current is given A trapezoidal pulse is generated according to the target magnetization state, and the direct axis current is set to a given value at other times. Set to 0; Will 、 As the input of the PI current controller, the voltage and current of the PI current controller output are decoupled, and after Park inverse transformation and space vector pulse width modulation, the driving signal is obtained and input into the inverter to drive the motor.

Images