Adaptive Parameter Adjustment of Differential Evolution

Abstract

In this paper, we focus on one of the Meta-Heuristics Differential Evolution (DE) and propose an adaptive parameter adjustment method to improve its search performance and usability. First, we define a scalar index based on numerical analysis according which diversification and intensification of search in DE can be executed. Moreover, we set an ideal target schedule for the index to reach a high level of search performance. Then we propose an adaptive parameter adjustment method to adjust the parameters properly so that the index can follow the ideal target schedule during the search process. Finally, we use several classic benchmark problems for the numerical experiment to verify the effectiveness of this newly proposed method.