基于自适应进化粒子群算法的多目标优化方法

Adaptive Evolutionary Particle Swarm Algorithm for Multi-Objective Optimisation

提出一种自适应进化粒子群优化算法(AEPSO),以提高多目标优化PSO算法的性能。AEPSO算法把非支配排序技术、自适应惯性权重和特殊的变异操作引入到PSO算法中,来提高算法的全局搜索能力和粒子的多样性。与常用的整体加权方法来处理多目标优化问题不同,AEPSO算法采用非劣解排序来引导粒子的飞行,以改进算法的收敛性,同时采用特殊的变异操作防止早熟收敛并增加优化解的多样性。所提算法的有效性经过四种代表性benchmark函数进行验证,并与几种典型同类型算法进行比较。该算法已成功地用于合金材料的多目标优化设计。实验结果表明AEPSO算法能够较好地兼顾收敛精度与优化解的多样性,满足多目标优化设计的要求。

An Adaptive Evolutionary Particle Swarm Optimisation （AEPSO） approach was proposed to improve the performance of PSO algorithm for multi-objective optimisation. The proposed approach incorporated non-dominated sorting, adaptive inertia weight and a special mutation operation into particle swarm optimisation to enhance the exploratory capability of the algorithm and improve the diversity of the Pareto solutions. Instead of using weighted aggregation approach to deal with multi-objective optimisation problems, dominance-based rank was used to guide the flight of particles. The proposed algorithm was validated using several well-known benchmark test functions and successfully applied to the multi-objective optimal design of alloy steels, which aimed at determining the optimal process parameters and the required weight percentages of the chemical composites in order to obtain the pre-defined mechanical properties of the materials. The experimental results have shown that the algorithm can locate the constrained optimal design with good accuracy while keep the diversity of the Pareto solutions.