Application of Response Surface Methodology and Elitist Multi-objective Hybrid Particle Swarm Algorithm for Optimization Design of an Air-Core Linear Motor

Abstract

The main purpose of this study was to find out the optimal design variables of an air-core linear brushless permanent magnet motor (LBPMM) by simultaneously considering the maximal thrust, minimal temperature, and minimal volume. Airgap length, magnet dimensions (magnet height and magnet width), and coil winding height were chosen as design variables in this multi-objective optimization problem. Using response surface methodology (RSM), this study developed a mathematical predictive model for each of the objectives. A multi-objective hybrid particle swarm optimization (PSO) with a mutation operator and a dynamic inertia weight factor was used to optimize the model developed by RSM. In addition, an elitist mechanism with crowding distance sorting was used to improve the correctness and diversity of the solutions. The results show that the Pareto-optimal front solutions of the proposed approach, providing the designers with more design plans, are superior to those of non-dominated sorting genetic algorithm ( NSGA II ). Designers can expand the effective method for designing linear motors to successfully solve more complex problems in other designed components.