一种基于模糊逻辑引入偏好信息的多目标遗传算法

Multi-objective Optimization Genetic Algorithm Incorporating Preference Information Based on Fuzzy Logic

针对多目标优化存在多个非支配解、用户难以挑选的问题,该文提出一种交互式引入决策者偏好信息的多目标遗传算法。该算法使用一种新型的九级标度赋值法把决策者通过语言表达的偏好信息量化为各目标的重要性因子,采用模糊推理系统构造一种基于偏好信息的'强度优于'关系替代常规的'Pareto支配'关系,以比较个体之间的优劣。对算法的计算复杂度进行了理论分析。仿真实验表明,该算法具有实时处理偏好信息的能力,与2种经典多目标遗传算法相比,该算法能够搜索到质量更优的解。

In order to solve the difficulty for users to select from many non-dominated solutions in multi-objective optimization,a multi-objective genetic algorithm incorporating preference information of the decision maker interactively is proposed.The algorithm makes use of a new nine-scale evaluation method to convert the linguistic preferences expressed by the decision maker to importance factors of objectives.A new outranking relation called ＂strength superior＂ which is based on the preference information is constructed via a fuzzy inference system to compare individuals instead of the commonly used ＂Pareto dominance＂ relation.The computational complexity of the algorithm is analyzed theoretically,and its ability to handle preference information is validated through simulation.Comparisons to two classical multi-objective genetic algorithms indicate that the proposed algorithm can search better solutions.