基于环境选择和配对选择的多目标粒子群算法

Multi-objective particle swarm optimization based on environmental selection and pairing selection

为了使多目标粒子群算法中种群粒子能够快速地收敛于帕累托最优边界,针对标准多目标粒子群算法中缺乏粒子评价标准以及种群个体历史最优值位置和全局最优值位置选择问题,提出了一种基于环境选择和配对选择策略的多目标粒子群算法。该算法在每次迭代时,采用SPEA2中的环境选择和配对选择策略及适应度值计算方法,以此来提高种群粒子之间的信息交换力度,减少标准多目标粒子群算法中大量的随机性,使种群粒子能够更快速地收敛于帕累托最优边界。经典测试函数的仿真实验结果表明,在标准多目标粒子群算法中运用SPEA2的环境选择、配对选择策略和适应度值计算方法,能够使种群粒子更快速地收敛于帕累托最优边界,验证了算法改进的可行性和有效性。

To make the population particle of PSO can more quickly converge to the Pareto optimal swarm,for the lacking of particle evaluation criteria and the weaker information exchange of the population particle for MOPSO,this paper proposed the environmental selection and pair selection strategy to MOPSO. When the algorithm was iterating,it used the environmental selection and pair selection strategy and the calculation method of fitness value in SPEA2,what depended to improve the efforts of information exchange between each particle populations,and reduced the randomness of standard MOPSO,what let the population particles could more quickly converge to the Pareto optimal boundary. Through the simulation of several classic test functions,the results show that the using of the environmental selection and pair selection strategy and the calculation method of fitness value of SPEA2 can let the population particle more quickly converge to Pareto optimal boundary in multi-objective particle swarm,what verify the feasibility and effectiveness of the improved algorithm.