Elite Strategy for Particle Swarm Optimization Algorithms
Elite strategy is proposed to improve the perfor-mance of Particle Swarm Optimization(PS0)algorithms.Several the best individuals from the best solution achieved so far by every particle fPbest)are selected to constitute an elite group at every iteration.When every particle updates its velocity.along every dimension the particle could jndependently select the nearest position from the same dimension of elite s as its attractor.Because elite s jointly guide the search for every particle instead of only one the best particle,elite strategy could effectively control the diversity of the swarm and then alleviate getting trapped into the local optimum.In our experiments.the classic PS0 algorithm,finearly decreasing weight particle swarm optimization(LDWPSO),is used to validate the effect of elite strategy.The globaIIocaI version of LDWPSO,the lOCal version of LDWPSO with von Neulnann topology.the iocaI version of LDWPS0 with circle topology,and LDWPSO equipped with dynamic elite strategy were compared.The experimental results show that elite strategy can greatly improve the performance of LDWPSO algorithms,especially,on complex function with many local minima.
Yu Liu Zheng Qin
国际会议
The International Conference Information Computing and Automation(2007国际信息计算与自动化会议)
成都
英文
673-677
2007-12-19（万方平台首次上网日期，不代表论文的发表时间）