基于组合排序的约束多目标优化算法-《计算机技术与发展》

文章信息/Info

Title:: Research on Constrained Multi-objective Optimization Algorithm Based on Combined Rank

Author(s):: LI Zhen-yu; HU Han; School of Computer Science and Technology,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China

Keywords:: constrained optimization; multi-objective optimization; constrained decomposition with grids; constraint handling

摘要:: 约束的多目标优化问题(CMOPs)常见于工程应用和现实生活中,这类问题往往包括多个冲突的目标以及一组约束条件。与无约束的多目标优化问题相比,此类问题包含了一些复杂的特征,解决起来也要困难得多。对此,文中提出了一种基于组合排序的约束处理方法。该方法与一个最新提出的基于约束分解的算法框架相结合来解决约束的多目标优化问题。基于网格的约束分解的进化算法(CDG-MOEA)是新提出的解决多目标优化的算法,在解决无约束多目标优化问题上具有多样性和鲁棒性等良好的特性。基于此框架,提出了基于组合排序的约束处理方法,旨在算法的每次进化中,选择出种群中多样性比较好且可行的那些解。为了验证算法的有效性,将提出的约束多目标优化算法(CDG-CS)与现有算法在多个约束的优化问题上进行实验分析,结果表明,该算法在约束的多目标优化问题上有着不错的效果。

Abstract:: Constrained multi-objective optimization problems (CMOPS) are common in engineering applications and real life,which often involve multiple conflicting objectives as well as a number of constraints. Compared with unconstrained many-objective optimization problems,CMOPs contain some complicated features and are much more difficult to solve. Therefore,we propose a constraint-handling mechanism based on a combined sorting,and combines it with a newly proposed optimization algorithm framework based on constrained decomposition to solve CMOPs. A constrained decomposition MOEA with grid (CDG-MOEA) is newly proposed to perform great in terms of diversity and robustness. Based on this framework,we propose the constraint-handling mechanism based on combined sorting,which aims to select solutions that are feasible and have better diversity. To verify the effectiveness of the algorithm, the proposed constrained multi-objective optimization algorithm (CDG-CS) will compare with other algorithms on some constrained optimization problems. The experiment demonstrates that CDG-CS can solve CMOPs very well.