An evolutionary multitasking optimization framework with fusion space for constrained multimodal multiobjective problems*

doi:10.1016/j.eswa.2026.132983

An evolutionary multitasking optimization framework with fusion space for constrained multimodal multiobjective problems*

2026-06-01

0

PRE

AI

DOI：10.1016/j.eswa.2026.132983

Original

Original request for help

Save

Abstract

En 中文

Constrained multimodal multiobjective optimization problems (CMMOPs) are challenging due to the difficulty in balancing convergence and diversity of solutions in the feasible regions. Although existing algorithms demonstrate good performance on most CMMOPs, their convergence and diversity are often suboptimal in CMMOPs with large scale infeasible region barriers. To overcome the above limitation, this work proposes an evolutionary multitasking optimization framework with a fusion space, called FSC-MTO, employing knowledge transfer from two auxiliary tasks to the main task to tackle such complex problems. In the proposed framework, the main task directly solves the original CMMOPs. The first auxiliary task employs constraint relaxation to expand the search space and guide the population toward potential regions. The second auxiliary task employs objective relaxation to enhance the diversity of population within the feasible region, preventing convergence to local Pareto solution sets. Furthermore, a fusion space-based selection mechanism is designed to simultaneously maintain the distribution of solutions in both the decision space and the objective space. Experimental validation on 31 benchmark problems and one practical case demonstrates that FSC-MTO significantly outperforms existing mainstream algorithms in solving complex CMMOPs.

Keywords:

Evolutionary algorithm

Constrained multimodal multiobjective

optimization

Multitasking

Fusion space

An evolutionary multitasking optimization framework with fusion space for constrained multimodal multiobjective problems*

Abstract

Journal

Organization

Cited Papers