Optimization and comparison of electrical machines using particle swarm optimization

Comparing different electrical machines is usually required at the start of a project. For a fair quantitative comparison the machines considered should be optimized under the given constraints, but optimally designing machines requires a significant amount of expertise and time. The aim of this paper is to present a flexible, numerical (2D) approach to optimize substantially different machines, thereby eliminating the complexities of designing a machine. The loss of insight into the problem caused by this approach is dealt with by using multi-target optimization and presenting the results as pareto optimal fronts. This allows a direct comparison of the machines under consideration and also enables the designer to immediately select a promising design for further analysis. The method is demonstrated on the design of three different PM machines.