摘要
Evolutionary algorithms are very popular for constrained multi-objective optimization problems (CMOPs). It is critical for constrained multi-objective evolutionary algorithms (CMOEAs) to pass across infeasible regions and acquire complex feasible fronts caused by constraints. In this paper, a constraint cone decomposition evolutionary algorithm (CCDEA) is proposed for CMOPs. CCDEA decomposes a CMOP into a series of constrained subproblems. And every subproblem is further decomposed into multiple constrained sub-layers. CCDEA maintains a decomposition-based population to preserve the best individual at each sub-layer. Thus CCDEA makes full use of the effective information of both feasible and infeasible solutions to pass across infeasible regions reasonably. In addition, a dominance-based population is designed in CCDEA to combine the advantages of decomposition and dominance to heighten the convergence and distribution of obtained fronts. Experimental results on DTLZ test instances with three types of constraints indicate that CCDEA achieves the overall better performances than the three other popular algorithms.
学科与关键词
- Cooperative Multiobjective Evolutionary Algorithm With Propulsive Population for Constrained Multiobjective OptimizationWang, Jiahai*; Li, Yanyue; Zhang, Qingfu; Zhang, Zizhen; Gao, ShangceIEEE Transactions on Systems Man Cybernetics-Systems, 2022, 52(6): 3476-3491.被引 20
- A Constrained Multi-Objective Evolutionary Algorithm Based on Boundary Search and ArchiveLiu Hai Lin; Peng Chaoda; Gu Fangqing*; Wen JiechangInternational Journal of Pattern Recognition and Artificial Intelligence, 2016, 30(1): 1659002.
- A Survey on Evolutionary Constrained Multiobjective OptimizationLiang, Jing; Ban, Xuanxuan; Yu, Kunjie*; Qu, Boyang; Qiao, Kangjia; Yue, Caitong; Chen, Ke; Tan, Kay ChenIEEE Transactions on Evolutionary Computation, 2023, 27(2): 201-221.被引 165
