CN-121984402-A - Model-free predictive control method for surface-mounted permanent magnet motor based on current gradient

CN121984402ACN 121984402 ACN121984402 ACN 121984402ACN-121984402-A

Abstract

The invention belongs to the technical field of motor predictive control, and in particular relates to a model-free predictive control method of a surface-mounted permanent magnet motor based on a current gradient, which comprises the steps of obtaining a current gradient equation according to a discretization result of a motor voltage equation; calculating the current gradient of each period, forming a lookup table, substituting each candidate voltage vector, obtaining a corresponding current gradient through the lookup table, calculating to obtain a current predicted value with delay compensation, selecting an optimal non-zero voltage vector by using a cost function, calculating the duty ratio of the optimal voltage vector, and introducing a zero voltage vector to realize double-vector model-free prediction current control. The invention solves the problem of current gradient update stagnation while ensuring the robustness of the model-free predictive control method, effectively improves the current prediction precision and improves the steady-state performance of speed and torque.

Inventors

LI XIANGLIN
Tian Diecen
DAI JUN
XU SHIQUN
GAO WEIWEI
LIU JIANZHE

Assignees

青岛大学

Dates

Publication Date: 20260505
Application Date: 20251215

Claims (6)

1. The model-free predictive control method for the surface-mounted permanent magnet motor based on the current gradient is characterized by comprising the following steps of: s1, establishing a current gradient equation of a motor driving system based on a surface-mounted permanent magnet motor mathematical model; S2, acquiring an operation parameter in a past control period to calculate a current gradient caused by an optimal non-zero voltage vector, and then calculating the current gradient caused by a candidate voltage vector to form a lookup table; S3, designing a cost function, applying a current predicted value to the cost function, and selecting a non-zero candidate voltage vector with a smaller cost function value as an optimal voltage vector; s4, calculating the optimal voltage vector duty ratio, and introducing a zero voltage vector to realize double-vector model-free predictive current control so as to obtain a corresponding power device switching signal for controlling the voltage source inverter.
2. The model-free predictive control method for a surface-mounted permanent magnet motor based on current gradients of claim 1, wherein the establishment of a current gradient equation of a motor driving system is specifically as follows: obtaining d-q axis voltage and current differential equation under a rotating coordinate system after coordinate transformation: (1); In the formula (1), V d 、V q is the d-q axis component of the stator voltage, i d 、i q is the d-q axis component of the stator current, R s is the stator resistance, omega r is the rotor electric angular velocity, psi f is the permanent magnet flux linkage amplitude, L d 、L q is the d-q axis inductance, and L d =L q =L s is satisfied for a surface-mounted permanent magnet motor; Obtaining a current gradient of a kth-1 period, namely a current gradient equation of a motor driving system, according to the discretization result of the formula (1): (2); In equation (2), i d k-1 and i q k-1 are the D-q axis components of the stator current in the k-1 period, respectively, i d k and i q k are the D-q axis components of the stator current in the k-1 period, respectively, T s is the sampling period, D k-1 is the duty cycle of the optimal non-zero voltage vector applied in the k-1 period, And Is the d-q axis component of the optimal non-zero voltage vector applied during the k-1 th period, And Is composed of And A current gradient respectively caused; The d-q axis component of the current gradient in the k-1 period is divided into two parts, namely a forced response component P 2 and P 3 which are influenced by the voltage vector, a natural response component P 1 and P 4 which are not influenced by the voltage vector, the forced response component is respectively proportional to the d-q axis component of the voltage vector, the natural response component is irrelevant to the voltage vector, and the d-q axis current gradient caused by the zero voltage vector is equivalently obtained.
3. The model-free predictive control method of a surface-mounted permanent magnet motor based on current gradients of claim 2, wherein the method comprises obtaining the operation parameters in the past control period to calculate the current gradients, then calculating the current gradients caused by the selected candidate voltage vectors from the obtained current gradients, and continuing to calculate the current gradients in the subsequent period to form a lookup table, and specifically comprises: acquiring an operation parameter in a past control period, calculating a current gradient caused by an optimal non-zero voltage vector in a formula (2), and controlling a first term of the optimization P1 according to a maximum torque current ratio, so that each part of a current gradient d-q axis component in a current gradient equation of the motor driving system is expressed as: (3); In the formula (3), S k-1 is d-axis current gradient forced response component generated by unit voltage amplitude in the kth-1 period, in the surface-mounted permanent magnet motor, d-axis and q-axis forced response components of current gradient caused by the unit voltage amplitude are equal, and sigma is a threshold value for judging S k-1 updating; in the multi-phase motor, the candidate voltage vector selects an actual voltage vector or a virtual voltage vector with the size of zero in an x-y subspace and the size of non-zero in an alpha-beta subspace, and the actual voltage vector or the virtual voltage vector is substituted into the candidate non-zero voltage vector to obtain a d-q axis component of a corresponding current gradient: (4); In the formula (4) of the present invention, And The d-q axis components of the current gradient caused by the candidate voltage vector with the sequence number j, i.e., VV j , in the k-1 th period, j being the sequence number of each candidate voltage vector, And The d-q axis components of the non-zero candidate voltage vector VV j during the k-1 period, respectively; And forming a lookup table by the calculated current gradients in the k-1 period, continuously calculating the current gradients in the subsequent period, and updating the current gradients to the lookup table.
4. The model-free predictive control method of a surface-mounted permanent magnet motor based on current gradients as set forth in claim 3, wherein substituting the selected candidate voltage vector to obtain the corresponding current gradient through a lookup table and calculating to obtain the current predictive value with delay compensation comprises Ignoring differences between current gradients caused by one voltage vector in adjacent control periods, replacing current gradients in k and k+1 periods with current gradients in k-1 period, i.e. Updating the current gradient caused by each candidate voltage vector for the present control period and the next control period using the sampling information from the previous control period; considering delay compensation, current prediction is performed by an updated lookup table: (5); in the formula (5), i s k is a current value in k cycles, i s k+1 、i s k+2 is current predicted values in k+1 and k+2 cycles, respectively, Is the current gradient caused by the optimal non-zero voltage vector applied during the k period, Is the current gradient caused by the zero voltage vector over the k period, D k is the duty cycle of the optimal non-zero voltage vector applied over the k period, Is the current gradient caused by the application of VV j during the k+1 period; , And Replaced by a corresponding current gradient in the k-1 period in the look-up table.
5. The model-free predictive control method for a surface-mounted permanent magnet motor based on current gradients of claim 4, wherein the cost function g of the design is: (6); in the formula (6) of the present invention, And Is a reference value of d-q axis current, and is set according to the actual running condition; And Is the d-q axis component of the current prediction value in the k+2 period.
6. The model-free predictive control method for a surface-mounted permanent magnet motor based on current gradients of claim 5, wherein the step S4 specifically comprises: the optimal non-zero voltage vector duty cycle in the k+1 period is calculated as follows: (7); In the formula (7) of the present invention, Is the q-axis component of the current gradient caused by the optimal non-zero voltage vector during the k-1 period; the optimal non-zero voltage vector action time is less than or equal to a control period, zero voltage vector is introduced for control in the remaining time, the selected zero voltage vector has a switching state of 000000, and a power device switching signal is obtained through PWM modulation by combining the optimal non-zero voltage vector and the corresponding duty ratio thereof and is used for controlling a voltage source inverter.

Description

Model-free predictive control method for surface-mounted permanent magnet motor based on current gradient Technical Field The invention belongs to the technical field of motor predictive control, and particularly relates to a model-free predictive control method for a surface-mounted permanent magnet motor based on a current gradient. Background In recent years, multiphase permanent magnet motors are widely focused in a plurality of high-end fields due to the advantages of excellent fault tolerance, flexible control strategies and the like, and have wide development prospects. At present, the control algorithm of the motor driving system mainly comprises vector control, direct torque control, model prediction control and the like, wherein the model prediction control has the characteristics of easiness in implementation and quick dynamic response, can directly and conveniently process various constraints, and is widely applied to the control of the permanent magnet synchronous motor driving system. The Chinese patent No. 119030399A discloses a permanent magnet synchronous motor model-free predictive current control method based on an expansion control set, which comprises the steps of obtaining related data sets of a three-phase permanent magnet synchronous motor, expanding original 7 basic voltage vectors in the model-free predictive current control method to 37, obtaining current gradients corresponding to the basic voltage vectors, obtaining k+1 control period predictive current after the voltage vectors act, obtaining a reference current gradient through delay compensation, determining quadrants of candidate voltage vectors according to the reference current gradient, reducing the area of the candidate voltage vectors, reducing the number of the candidate voltage vectors from 37 to 6, calculating to obtain corresponding current gradients according to the candidate voltage vectors V xy, selecting an optimal voltage vector from the candidate voltage vectors according to a cost function, and generating an inverter control signal. However, model predictive control has the defect of severely depending on motor parameter accuracy, so model-free predictive control based on current gradients is provided. However, this method has a problem of update stagnation, which results in significant errors in current prediction. In addition, if only one voltage vector is used in one period, that is, single-vector control is performed, a large current ripple is caused, but the existing gradient update method is not suitable for multi-vector model-free prediction control, and satisfactory control accuracy is difficult to achieve. In addition, the traditional model-free predictive control objects are three-phase permanent magnet synchronous motors, and research on the multiphase permanent magnet motors is lacking. Disclosure of Invention Aiming at the problems, the invention aims to provide a model-free predictive control method for a surface-mounted permanent magnet motor based on current gradients, which is suitable for three-phase and multi-phase permanent magnet motors of various types, performs current prediction based on current gradients generated by candidate voltage vectors, does not need to depend on motor parameters, only uses sampling information from a previous control period, and realizes accurate control on a motor driving system through double-vector model-free predictive control so as to solve the problem of update stagnation existing in the current model-free predictive control based on the current gradients and reduce torque and rotation speed pulsation when the motor runs. The detailed technical scheme of the invention is as follows: A model-free predictive control method for a surface-mounted permanent magnet motor based on current gradients comprises the following steps: s1, establishing a current gradient equation of a motor driving system based on a surface-mounted permanent magnet motor mathematical model; S2, acquiring an operation parameter in a past control period to calculate a current gradient caused by an optimal non-zero voltage vector, and then calculating the current gradient caused by a candidate voltage vector to form a lookup table; S3, designing a cost function, applying a current predicted value to the cost function, and selecting a non-zero candidate voltage vector with a smaller cost function value as an optimal voltage vector; s4, calculating the optimal voltage vector duty ratio, and introducing a zero voltage vector to realize double-vector model-free predictive current control so as to obtain a corresponding power device switching signal for controlling the voltage source inverter. Further, the establishment of the current gradient equation of the motor driving system is specifically as follows: obtaining d-q axis voltage and current differential equation under a rotating coordinate system through coordinate transformation: (1); In the formula (1), V d、Vq is the d-q axis component