A Genetic Algorithm with Tabu List and Sharing Scheme for Optimal Design of Electrical Machines

This paper presents a new genetic algorithm with tabu list and sharing scheme for optimizing the design of electrical machines. The design problem was formulated as a multimodal function optimization problem. All the obtained solutions are divided into clusters before being judged by a trained neural network to decide which solution is optimal. From the results of examples, as well as comparisons with classical genetic algorithms (GAs) with sharing scheme, we can see that the proposed GA reveals the intrinsic solution region distribution more distinctly, therefore having more possibility to Žnd out the global optimal solution for design of electrical machines.