A global optimal technique based on Moving Least Square and Improved Differential Evolution

This paper presents a global optimal technique based on moving least square (MLS) fitting technique and improved differential evolution algorithm (IDEA) for inverse problems on optimizing electromagnetic equipments. In the proposed method, MLS are firstly used to simulate complex objective functions as a response surface model (RSM) in multidimensional space, for transforming an implicit function to an explicit one, then differential evolution algorithm (DEA) is improved and used by combining with MLS to get the global optimization with high accuracy and efficiency. TEAM Workshop Problem 22, as numerical benchmark, indicates that the proposed method is superior to other random optimal techniques, and it can be comprehensively used for inverse problems in engineering electromagnetic applications.