An improved sliding mode observer algorithm for PMSM based on deformable fuzzy neural network

To address the issue of poor speed estimation performance caused by the inability of sensor parameters to adjust in real time under complex working conditions in traditional sliding mode observer sensorless control techniques for permanent magnet synchronous motors (PMSM), an improved sliding mode sensorless algorithm combining radial basis function (RBF) neural networks and fuzzy logic control theory is proposed. This improved algorithm accurately obtains the Jacobian matrix of the PMSM through an RBF neural network parameter identifier, and based on this, it rapidly determines the structural situation of the deformable fuzzy neural network through a variable structure learning process. Simulation results obtained from MATLAB/Simulink demonstrate that this estimation algorithm enhances the speed control accuracy of the PMSM vector control system by 5.9% when compared to the proportion integral differential (PID) speed controller algorithm and by 26.9% when compared to the sliding mode speed controller algorithm. The findings suggest that the proposed algorithm enhances the accuracy of speed estimation and rotor position while optimizing the speed control performance of vector control system.