一种基于和声搜索的动态交叉粒子群算法

Dynamic Crossover Particle Swarm Optimization Based on Harmony Search

针对粒子群优化算法(PSO)在优化多维问题时容易陷入局部最优的问题,提高其全局搜索能力和拓展能力,提出了一种基于和声搜索的动态交叉粒子群算法.引入动态交叉操作,使得粒子在更新速度时实现共享有效信息,保证粒子进化过程中的种群多样性,提高全局搜索能力.结合和声搜索(HS)的随机搜索能力提出了HS-DCPSO,利用和声搜索的自适应调整参数音符调节概率PAR和间隔调整带宽bw来提高粒子群的拓展能力.通过多个基准函数对所提出的HS-DCPSO算法进行仿真测试,并与HS、PSO及多种改进的粒子群算法对比,验证所提出的HS-DCPSO算法具有较强的全局搜索能力和局部拓展能力,并且算法时间复杂度相比传统PSO增加不明显.

Aiming at the local trap problem of particle swarm optimization （ PS0 }, this paper is supposed to improve its global search and exploit ability. We carried out a dynamic crossover PSO based on Harmony Search. Dynamic crossover operation is adopted to make particles share useful information among each other which would guarantee both population diversity and global search ability. HS-DCPSO is brought up combined with random search ability of harmony search. Parameter Pitch adjusting rate { Par } and arbitrary distance bandwidth bw would be self-adaptive to improve exploit ability of particles. HS-DCPSO are testified on various benchmark functions and compared with HS, PSO and various improved PSOs. The simulation results proved that HS-DCPSO has better per- formances of global search and exploit ability without obvious algorithm complexity incensement.