Design Optimization of a Switched Reluctance Machine using Analytical model and PSO

Abstract

This paper proposes a fast and efficient process to optimize design variables of a switched reluctance machine (SRM). The major objective of this optimization problem is reduction of torque pulsation. Therefore, design variables are selected based on their sensitivities towards torque ripple. As chosen variables also affect output power, it leads to multiobjective optimization, where maximization of output power is considered as second objective. However, it is difficult to express both the objectives as explicit mathematical functions of chosen design variables. The aforementioned problem necessitates an analytical model of an SRM which computes data as per objective function for given design variables. Particle swarm optimization technique is used to determine optimum values of design variables. For validation, at first, an initial design is carried out using the conventional method. Then, optimized design variables are obtained using the proposed process. Finally, performances of the initial and optimized designs, obtained from finite element method (FEM) based ANSYS Maxwell platform, are compared. Also, this paper compares simulation times required for proposed optimization process and data generation using FEM in conventional process.