CN-122026765-A - Permanent magnet synchronous motor current control method and device and electronic equipment

CN122026765ACN 122026765 ACN122026765 ACN 122026765ACN-122026765-A

Abstract

The invention discloses a permanent magnet synchronous motor current control method and device and electronic equipment. The method comprises the steps of calculating deviation between a current voltage modulation degree and a preset upper limit, generating virtual rotation speed increment through adjustment, superposing the virtual rotation speed increment to an actual rotation speed to obtain an equivalent table lookup rotation speed, inquiring a static association relation table by the equivalent table lookup rotation speed to obtain a reference current instruction so as to realize anti-saturation control, collecting historical running state data of a motor, extracting first-order and second-order change rates, mapping the original data and the change rates into a two-dimensional feature matrix, inputting the matrix into a pretrained convolutional neural network to extract dynamic trend features, outputting dynamic compensation current, and finally superposing the reference current instruction and the compensation current to control the motor. According to the invention, voltage safety is ensured through the virtual rotating speed main channel, dynamic errors and parameter drift are eliminated through the second-order trend imaging compensation channel, and optimal current control under all working conditions is realized.

Inventors

LI WUHUA
ZHANG MENGGUANG
DOU YU
SHENG JING
Yuan Chongzhe
LIU YING

Assignees

杭州市拱墅区全息智能技术研究院

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (7)

1. A method for controlling current of a permanent magnet synchronous motor, comprising: acquiring current running state data and a torque command of a motor; calculating the current voltage modulation degree, and calculating the deviation between the voltage modulation degree and the upper limit of a preset modulation degree; Adjusting the deviation to generate a virtual rotation speed increment, and superposing the virtual rotation speed increment to the actual rotation speed of the motor to obtain an equivalent table look-up rotation speed; Inquiring a pre-stored static association relation table by using the equivalent table lookup rotating speed and the torque command to obtain a reference current command; collecting running state data in a preset historical time window to construct a state sequence, and performing differential calculation on the state sequence to extract a first-order change rate and a second-order change rate; the original data, the first-order change rate and the second-order change rate in the state sequence are arranged according to time steps and mapped into a two-dimensional feature matrix; inputting the two-dimensional feature matrix into a pre-trained convolutional neural network, extracting dynamic trend features through a convolutional kernel, and outputting dynamic compensation current; and superposing the reference current instruction and the dynamic compensation current to obtain a final current instruction, and inputting the final current instruction into a current loop controller to control the operation of the motor.
2. The method of claim 1, wherein the operating state data includes d-axis current error, q-axis current error, rotational speed, bus voltage, and temperature; the construction process of the two-dimensional feature matrix comprises the following steps: taking the running state data as zero-order data; calculating the difference value of the running state data in adjacent time steps as a first-order change rate; Calculating the difference value of the first-order change rate in adjacent time steps as a second-order change rate; And stacking the zero-order data, the first-order change rate and the second-order change rate in the vertical axis direction, and expanding the stack according to time steps in the horizontal axis direction to generate the two-dimensional feature matrix.
3. The method of claim 1, wherein the convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, and a fully-connected layer, connected in sequence; The convolution layer is used for extracting local dynamic texture features in the two-dimensional feature matrix; the full connection layer is used for mapping the extracted characteristics into d-axis current compensation values and q-axis current compensation values; The training objective function of the convolutional neural network comprises a torque tracking error term and a system efficiency reciprocal term.
4. The method of claim 1, wherein the step of adjusting the deviation to generate a virtual rotational speed delta comprises: the deviation is regulated by utilizing a proportional integral controller, and a frequency increment is output; converting the frequency increment into a rotating speed unit to obtain the virtual rotating speed increment; And when the voltage modulation degree exceeds the upper limit of the preset modulation degree, the virtual rotating speed increment is a positive value, so that the equivalent table look-up rotating speed is larger than the actual rotating speed.
5. A permanent magnet synchronous motor current control device, characterized by comprising: the data acquisition module is used for acquiring current running state data and torque instructions of the motor; The main channel control module is used for calculating the deviation between the current voltage modulation degree and the upper limit of the preset modulation degree, generating a virtual rotation speed increment, superposing the virtual rotation speed increment to the actual rotation speed to obtain an equivalent table lookup rotation speed, and inquiring a static association relation table by using the equivalent table lookup rotation speed to obtain a reference current instruction; The characteristic construction module is used for collecting historical running state data to construct a state sequence, extracting a first-order change rate and a second-order change rate, and mapping the original data, the first-order change rate and the second-order change rate into a two-dimensional characteristic matrix; The prediction compensation module is used for inputting the two-dimensional characteristic matrix into a pre-trained convolutional neural network and outputting dynamic compensation current; and the instruction synthesis module is used for superposing the reference current instruction and the dynamic compensation current to obtain a final current instruction.
6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 4 when the program is executed by the processor.
7. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any one of claims 1 to 4.

Description

Permanent magnet synchronous motor current control method and device and electronic equipment Technical Field The invention relates to the technical field of electric drive control of new energy automobiles, in particular to a Permanent Magnet Synchronous Motor (PMSM) current loop anti-saturation control method and device and electronic equipment combining modulation degree closed-loop feedback and deep learning prediction compensation. Background The power system core of the new energy automobile is an electric drive system, wherein a Permanent Magnet Synchronous Motor (PMSM) is widely used by virtue of its high power density and high efficiency. In a vector control (FOC) system of a motor, a current loop is used as a control loop of an innermost layer and is responsible for precisely controlling d-axis and q-axis currents of the motor according to a torque command. In conventional control logic, the current loop calculates a voltage command from the current error through a PI (proportional integral) regulator and sends it to a Space Vector Pulse Width Modulation (SVPWM) module to drive the inverter. However, the output voltage capability of the inverter is limited by the dc bus voltage. When the vehicle is running at high speed (high back emf) or the driver is stepping on the throttle (sudden torque command increases cause a dramatic change in q-axis current command), the voltage command calculated by the PI regulator often exceeds the maximum voltage oval limit that the inverter can provide. At this point, the system enters a "voltage saturated" state. Once saturation occurs, the actual voltage cannot track the command voltage, resulting in a slow actual current response and a failure to track the current command in time. More seriously, the integral term of the PI controller continues to accumulate (integral saturation/windup), resulting in a severe overshoot. When the system attempts to exit saturation, the regulation process is slow due to the excessive integral term, which can cause severe oscillations in current and torque. On the whole vehicle level, the method is characterized in that the vehicle is in weak acceleration and delayed in response, even violent play or jerk appears, and the driving experience and the driving safety are seriously affected. Existing anti-saturation schemes typically focus on clamping of PI integral terms or complex flux weakening algorithms, but these methods tend to be computationally complex and have insufficiently smooth transitions under dynamic conditions. Therefore, there is a need for an optimized solution that can actively, quickly and smoothly prevent current loop saturation. Disclosure of Invention The invention aims to provide a method, a device and electronic equipment for controlling current of a permanent magnet synchronous motor, wherein a modulation degree is introduced as a feedback variable, and a table look-up working point is actively adjusted when a system is about to enter a saturation region, so that current loop saturation is avoided, and system stability is improved. In order to achieve the above object, the present invention provides a current control method for a permanent magnet synchronous motor, including: acquiring current running state data and a torque command of a motor; calculating the current voltage modulation degree, and calculating the deviation between the voltage modulation degree and the upper limit of a preset modulation degree; The deviation is regulated to generate a virtual rotating speed increment, and the virtual rotating speed increment is overlapped to the actual rotating speed of the motor to obtain an equivalent table look-up rotating speed; Inquiring a pre-stored static association relation table by using the equivalent table lookup rotating speed and the torque command to obtain a reference current command; collecting running state data in a preset historical time window to construct a state sequence, and performing differential calculation on the state sequence to extract a first-order change rate and a second-order change rate; the original data, the first-order change rate and the second-order change rate in the state sequence are arranged according to time steps and mapped into a two-dimensional feature matrix; inputting the two-dimensional feature matrix into a pre-trained convolutional neural network, extracting dynamic trend features through a convolutional kernel, and outputting dynamic compensation current; And superposing the reference current command and the dynamic compensation current to obtain a final current command, and inputting the final current command into a current loop controller to control the operation of the motor. Preferably, the operation state data includes d-axis current error, q-axis current error, rotation speed, bus voltage and temperature; The construction process of the two-dimensional feature matrix comprises the following steps: taking the running state data as zero-order data; cal