Methods to balance the exploration and exploitation in Differential Evolution from different scales: A survey

Abstract

Inspired by the evolutionary process in nature, Differential Evolution (DE) has been widely concerned and used as a numerical global optimizer for decades of years, since its emerging in 1997. However, the performance of DE essentially depends on the balance of its exploration ability and exploitation ability. To better summarize the recent works on DE, especially from 2019 to 2023, this paper analysed the balancing strategies from different scales, including from the algorithm level, the operator level and the parameter level. And then, all of the recent works are categorized and discussed according to different scales. From the algorithm level, the hybridizing methods of DE are mainly reviewed. For the evolution operators, both the enhanced operators and operator selection strategies are introduced. And for the parameters of DE, mainly different adaptation controlling strategies are summarized. The main purpose of this paper is to give an update summary of DE research and review these works on exploration–exploitation balancing from different scales.