摘要
为改善和声搜索算法易陷入局部最优的不足,提出了一种混沌反向学习和声搜索(COLHS)算法.基于聚集和发散思想,对算法陷入局部最优和停滞状态进行初步预判断,并根据预判断的结果融合混沌扰动策略和反向学习,利用了logistic混沌序列的遍历性和反向学习的空间可扩展性.此外,利用和声记忆库的历史信息定义更新因子和进化因子,自适应地调整参数基音调整概率(PAR)和基音调整步长(BW),平衡算法的聚集和发散.数值结果表明,COLHS算法优于HS算法及最近文献报道的8种改进的HS算法.
Harmony search(HS) algorithm is easily trapped into local optimal.To improve this shortcoming,chaos opposition-based learning harmony search(COLHS) algorithm w as proposed.Based on the thought of aggregation and divergence,preliminary judgments w hether this algorithm w as trapped into local optimal or backw ater status w ere given,then according to the judge result,disturbance strategy w as integrated w ith opposition-based learning technology.The ergodicity of logistic chaos sequence and the space extensibility of opposition-based learning w ere used.Besides,to balance aggregation and divergence,the history information of harmony memory w as used to define the updating factor and the evolution factor,w hich w ere applied to dynamically adjust the pitch adjustment rate(PAR) and the bandw idth(BW).Numerical results demonstrated that the proposed algorithm is better than HS and the other eight kinds of improved HS algorithms that reported in recent literatures.
出处
《东北大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第9期1217-1221,共5页
Journal of Northeastern University(Natural Science)
基金
国家自然科学基金资助项目(81000639)
关键词
和声搜索算法
混沌扰动策略
反向学习
局部最优
历史信息
harmony search algorithm
chaos disturbance strategy
opposition-based learning
local optimal
history information