Solving dynamic multi-objective engineering design problems via fuzzy c-means prediction algorithm

• Published in: Swarm and evolutionary computation Vol. 98; p. 102057
• Main Authors: Zhang, Qingyang, Fu, Xueliang, Yang, Shengxiang, Jiang, Shouyong, Li, Miqing, Zheng, Zedong
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.10.2025
• Subjects: Dynamic multi-objective optimization; Fuzzy C-means algorithm; Partial mutation strategy; Regression analysis; Regression analysis; Dynamic multi-objective optimization; Partial mutation strategy; Fuzzy C-means algorithm
• ISSN: 2210-6502
• DOI: 10.1016/j.swevo.2025.102057

This paper proposes a new prediction algorithm by integrating the fuzzy c-means and regression analysis fitting techniques with multi-objective differential evolution (FRMODE) to solve dynamic multi-objective optimization problems. When environmental changes are detected, the main purpose of FRMODE is to predict high-quality populations that can effectively track the moving Pareto-optimal set. Specifically, the fuzzy c-means (FCM) algorithm clusters the populations obtained from the past two adjacent environments. The center points of populations are utilized to define the moving direction, which is used to predict high-quality agents based on previous non-dominated individuals. Then, linear and non-linear regression analysis fitting strategies are developed to model the distribution of variables according to the variables’ characteristics. Besides that, the partial mutation strategy is also utilized to guide individuals toward more promising regions by intensifying the search around current agents. To evaluate the performance of the proposed algorithm, experiments are conducted on a set of benchmark functions with various dynamic difficulties, as well as on two classical dynamic engineering design problems. The experimental results demonstrate that FRMODE is more competitive compared with several state-of-the-art algorithms.