一种求解鲁棒优化问题的多目标进化方法

鲁棒优化问题(Robust Optimization Problem,ROP)是进化算法(Evolutionary Algorithms,EAs)研究的重要方面之一,对于许多实际工程优化问题,通常需要得到鲁棒最优解。利用多目标优化中的Pareto思想优化ROP的鲁棒性和最优性,将ROP转化为一个两目标的优化问题,一个目标为解的鲁棒性,一个目标为解的最优性。针对ROP与多目标优化的特点,利用动态加权思想,设计一种求解ROP的多目标进化算法。通过测试函数的实验仿真,验证了该方法的有效性。

Evolutionary multi-objective approach for solving robust optimization problem

Robust Optimization Problem（ROP） is one of the most important parts of Evolutionary Multiobjective Optimiza-tion（EMO）.For most practical engineering optimization problems,the aim of them is to obtain robust optimal solutions.In this paper,the concept of Pareto in multiobjective optimization is employed to deal simultaneously with robustness and opti-mality.Therefore,a ROP is transformed into a biobjective problem,one of which is the robustness of solution and the other is the optimality of solution.Combining the characteristics of ROP and multi-objective optimization,a Multi-Objective Evolu-tionary Algorithm（MOEA） for solving ROPs is designed by dynamic weight strategy.By the several experiments on two ROP test problems,the results demonstrate that the proposed evolutionary multi-objective approach is efficient.