一种具有初始化功能的自适应惯性权重粒子群算法-信息与控制.PDF

41 2 Vol.41, No.2 2012 04 Information and Control Apr., 2012 DOI 10.3724SP.J.1219.2012.00165 1 1 2 1 1. 710038 2. 95961 100195 PSO PSO PSO LDW-PSO TP301 A 1002-0411(2012)-02-0165-05 A Particle Swarm Optimization Algorithm with Initialized Adaptive Inertia Weights DU Jiyong1 ZHANG Fengming1 LI Jianwen2 YANG Ji1 (1. Engineering College, Air Force Engineering University , Xi’an 710038, China; 2. Unit 95961 of PLA, Beijing 100195, China) Abstract: An improved particle swarm optimization (PSO) algorithm with initialized adaptive inertia weights is proposed. An adaptive inertia weight strategy is introduced into the algorithm to balance the global and the local search abilities. In the evolution process, the swarm trapping at the stagnation is reinitialized around its weighted centroid position, which guides the swarm to avoid local extremum and accelerates the convergence rate of the algorithm. The comparison among the performances of the proposed approach, the standard PSO and the LDW-PSO is done. The experiment results show that the proposed algorithm can not only enhance the ability of avoiding local extremum, but also speed up the convergence rate and improve the stability to a certain degree. Keywords: particle swarm optimization algorithm; adaptive; inertia weight; initialization; weighted centroid 1 Introduction inertia weight [4-5] [3] [6] PSO