Torque modeling of Switched Reluctance Motor using LSSVM-DE

Abstract

An accurate and precised model of Switched Reluctance Motor (SRM) captures the performance of the machine. The model is highly nonlinear and complicated due to its doubly salient structure. In this paper, a nonparametric regression model for torque parameter of SRM was developed using Least Square Support Vector Machine (LSSVM) applying the concept of kernel based learning. Hyperparameters of LSSVM improves the model accuracy and generalization ability. The optimal values of hyperparameters were obtained by using Differential Evolution (DE) and high accuracy was observed. The regression model was developed using the magnetic characteristics of SRM by 2-D Finite Element Methods (FEM). The results obtained have been compared with the other optimization algorithms such as Genetic Algorithm (GA) and Particle Swarm Optimization (PSO).