一种新的求解多目标问题的分组粒子群优化算法

A New Divisional PSO Algorithm for Solving Multi-objective Optimization

用粒子群优化算法求解多目标问题容易陷入局部最优,为此本文提出了一种分组粒子群多目标优化算法。该算法将决策空间分成Q个子空间,每个子空间随机的分配N个粒子,这Q个粒子群分别在各自的空间进行独立搜索。为保证每个种群的搜索多样性和遍历性,用混沌序列对各组粒子位置进行初始化,同时对各组进行基于聚集距离的粒子择优进化。由典型多目标函数的优化实验结果表明,经过适当的分组,该算法能迅速逼近非劣最优解集,效果令人满意。

In order to solve the problem that it is easily plunged into local optima to use particle swarm optimization （ PSO） al-gorithm for multi-objective problem, this paper proposes a divisional PSO algorithm, named MODPSO.This algorithm divide func-tion domain into Q subspaces, each subspace will be randomly allocated N particles.These Q particle swarm search independently in their own space respectively.In order to guarantee each species＆#39;diversity and ergodicity of searching, chaotic sequence and crowding distance is used to initiate individual position and select the best individual .By proper dividing, experimental results on several typical multi-objective function show that the algorithm can rapidly find the Pareto optimal which is quite satisfactory .