Study on the sensorless control method of permanent magnet synchronous motor based on self-tuned boundary layer and RBF neural network
This study develops a proportional-integral-derivative (PID) control framework synthesizing self-tuning boundary layer sliding mode observers with radial basis function (RBF) neural networks. This approach enables adaptive adjustment of the velocity loop PI controller and boundary layer self-tuning...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2025-09-01
|
| Series: | Results in Engineering |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2590123025018870 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study develops a proportional-integral-derivative (PID) control framework synthesizing self-tuning boundary layer sliding mode observers with radial basis function (RBF) neural networks. This approach enables adaptive adjustment of the velocity loop PI controller and boundary layer self-tuning in the sliding mode observer, dynamically optimizing control parameters under varying operating conditions. It effectively addresses the challenge of balancing control accuracy and chatter suppression when using fixed boundary layer parameters. This approach facilitates precise monitoring of permanent magnet synchronous motor (PMSM) rotor position and rotational speed, concurrently mitigating chattering phenomena. To reduce chattering, the hyperbolic tangent function is employed to replace the sign function. The systematic analysis of the effect of boundary layer gain on sliding mode observers is presented. The relationship between boundary layer gain and relevant parameters is summarized, and selection criteria for boundary layer gain are established to balance control accuracy with chattering suppression. To realize the self-tuning functionality of boundary layer gain, an adaptive adjustment mechanism based on the derivative of current error is designed, enabling dynamic optimization of the sliding mode observer's boundary layer. Furthermore, an RBF neural network captures the system's Jacobian information, enabling real-time adaptive tuning of velocity-loop PID parameters. A variable-speed integration strategy is introduced to enhance the integral component and mitigate integral saturation phenomena. Given the consistent agreement between simulation outcomes and experimental data, the novel strategy proves operationally feasible and functionally superior. |
|---|---|
| ISSN: | 2590-1230 |