Optimized electric machine design solutions with efficient handling of constraints
Abstract
Design optimization of electric machines often requires time-expensive Finite Element Analysis (FEA) to evaluate objective functions accurately. In contrast to time-consuming simulations, the constraints of a feasible design can be based on relatively simple geometric expressions, resulting in a mixed computationally expensive, constrained multi-objective optimization problem. The proposed optimization method incorporates a customized repair operator into the well-known evolutionary multi-objective optimization algorithm NSGA-II. The repair operator exploits the low computational cost of constraints to ensure the feasibility of all solutions. The effectiveness of the proposed method is demonstrated by comparing results from two different optimization approaches. Apart from the optimization method itself, a physical explanation of Pareto-optimal solutions and insights about electric machine design are also discussed. Finally, a posteriori approach for selecting preferred solutions from the non-dominated set is presented while highlighting the trade-off offered by selected solutions.
Authors: Bhuvan Khoshoo, Julian Blank, Thang Q. Pham, Kalyanmoy Deb, Shanelle N. Foster
Published in: IEEE Symposium Series on Computational Intelligence (SSCI) (2021)