Comparative SQP-GA-PSO Algorithms for Hierarchical Multi-Objective Optimization Design of Induction Motors

This paper presents the optimal design for a 30 kW, 3-phase squirrel-cage induction motor (IM). In this paper, three optimization algorithms are used for design optimization, namely, Particle Swarm Optimization Algorithm (PSO), genetic algorithm (GA), and Sequential Quadratic Programming (SQP). The optimal goals are maximum starting torque, efficiency, and minimum material cost. The result of the IM design optimization using three optimal methods is announced and compared. Additionally, computation time and the number of iterations of each algorithm are compared to find out the most suitable algorithm for the optimal design of an induction motor. In addition, this paper proposes a solution that permits us to find only one solution satisfying all the optimal criteria. Instead of using the conventional multi-objective optimization method that normally leads to a Pareto set with many optimal points at the same optimal level, we propose a hierarchical optimization method that experiences some mono-objective optimization and then builds a function representing the multi-objective optimization. Using this method, having a global optimal point can be obtained. Comparison of the optimal algorithms and multi-objective optimization methods has given broadened insight into optimal techniques for IMs. We have found that PSO is the best method for optimization design of IMs in terms of computation time and finding the global optimal point.