Optimization of Single Pulse Control based on Design of Experiments for Switched Reluctance Machines

Abstract

Owing to the robust structure of switched reluctance machines, high-speed operations are preferred to achieve higher power densities. Generally, at high speeds, single pulse control is applied. However, its control parameter set is not unique. This paper proposes an algorithm based on design of experiments that simultaneously allows the calculation of optimal control parameters for an operating area and the minimization of the total machine losses. The designed experiments are finite-element analyses (FEAs) considering both iron and copper losses. A Pareto search algorithm solves the proposed multi-objective optimization problem. The results for different response surfaces are compared considering the errors in torque production and machine loss prediction. A reduced response surface based on 15 FE-runs shows a mean error of 1.9%, whereas the maximum error is 6.3% for machine losses within the validation operating area.