Position Sensorless Control of SRMs Based on Neural Network Nonlinear Modeling Using Improved Coyote Optimization Algorithm

This article presents a sensorless control method of switched reluctance motor (SRM) based on flux linkage nonlinear modeling. This method combines radial basis function neural network () with improved coyote optimization algorithm (ICOA) to obtain accurate rotor position information. The radial basis function neural network (RBFNN) has fast convergence speed and strong approximation ability. The ICOA can dynamically adjust the parameters and structure of RBFNN, and reduce modeling errors by adaptive adjustment of the number of hidden layer nodes. In addition, the rotation speed of the rotor at any position is estimated to realize the accurate commutation of the motor and ensure the stable operation of the whole closed-loop system. Finally, the effectiveness of the RBFNN based on the ICOA is verified through the six-phase 12/10SRM experimental platform.