.png){kind=icon}
.png){kind=icon}
Return
Dual-population hybrid prediction co-evolutionary algorithm for dynamic constrained multiobjective optimization
.png){kind=icon}
PRE
AI
DOI：10.1016/j.eswa.2026.131649.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Dynamic constrained multi-objective optimization problems (DCMOPs) are common in the real world, where maintaining the diversity and feasibility of the search population is crucial for such problems, especially in high uncertainty conditions. To tackle these challenges, this paper proposes a dual-population hybrid prediction co-evolutionary algorithm (DPHP). Specifically, the main population is dedicated to searching the feasible region, while the auxiliary population addresses an unconstrained (M+1)-objective problem that incorporates both the original M objectives and the degree of constraint violation, possessing the ability of global exploration. To improve the adaptability under high uncertainty, a hybrid prediction strategy is adopted: the main population employs a center-based prediction to process local search after environmental changes, whereas the auxiliary population utilizes a grey model-based clustering prediction to enhance the diversity search between the objective space and the constraint space. In addition, a dual-archive mechanism is employed to preserve promising solutions from both populations, providing valuable guidance for the main population. Extensive experiments on benchmark problems and a real-world problem against six state-of-the-art algorithms demonstrate that DPHP consistently achieves superior diversity preservation and adaptability.
Keywords:
Dynamic constrained multiobjective
Evolutionary algorithm
Hybrid prediction
Dual population
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W