Diversity-based adaptive differential evolution algorithm for multimodal optimization problems

doi:10.1016/j.swevo.2025.101869

Return

Diversity-based adaptive differential evolution algorithm for multimodal optimization problems

2025-03-01

PRE

DOI：10.1016/j.swevo.2025.101869

Original

Original request for help

Save

Abstract

En 中文

Multimodal optimization problems (MMOPs) require algorithms to locate as many global optima as possible. In recent years, evolutionary algorithms (EAs) have been widely employed to address MMOPs. A fundamental requirement when utilizing EAs for MMOPs is to maintain a delicate balance between global exploration and local exploitation, as well as to ensure that the algorithm can effectively escape local optima. In this paper, we propose a diversity-based adaptive differential evolution (DADE) algorithm to deal with these challenges. First, a diversity-based niching method not sensitive to the choice of parameters is proposed. This method can divide the population into several niches with appropriate sizes at different search stages, thereby enabling the algorithm to thoroughly explore the entire fitness landscape. Second, a mutation selection strategy with diversity control is devised to enable each niche to adaptively choose an appropriate mutation scheme at each iteration, so as to allow each subpopulation to better balance diversity and convergence. Third, to enable the individuals within a prematurely convergent subpopulation to escape local optima and subsequently perform a more effective search, a novel local optima processing strategy based on a carefully designed archive is introduced. Experimental results on 20 multimodal benchmark functions and a real-world application demonstrate that DADE exhibits greater robustness across diverse landscapes and dimensions compared to several state-of-the-art competitors, and also showcasing strong effectiveness and practicality. The code for DADE is available at https://github.com/qianxian -sheng/DADE.git.

Keywords:

Multimodal optimization problems (MMOPs)

Differential evolution (DE)

Niching

Population diversity

Local optima processing