An Improved Differential Evolution Algorithm Adopting $\lambda$ -Best Mutation Strategy for Global Optimization of Electromagnetic Devices

The differential evolution (DE) algorithm has almost ten different variants according to a trial vector generation strategy. The trial vector generation strategy has a significant effect on the performance of the DE algorithm. The selection of a suitable mutation strategy, however, is difficult because of the differences in the convergence speed and diversity. This paper proposes an improved differential evolution algorithm adopting a new mutation strategy, “DE/λ-best/1,” to increase the performance of global optimization. The suggested mutation strategy guides the population to the feasible region of various constraint optimization problems. The validity and numerical efficiency of the developed method was investigated through a comparison with conventional DEs on well known benchmark functions and Testing Electromagnetic Analysis Methods (TEAM) problem 22.