基本和声搜索算法的全局搜索能力较强,但局部收敛速度较慢,针对该不足,提出了一种改进方案.在基本和声搜索算法的基础上,采用分阶段优化的思想,在算法迭代的不同阶段分别采取不同的策略改进算法的局部收敛速度.在迭代初期采用混沌策略初始化和声库,在迭代的中后期采用聚类和混沌扰动策略进行加速,在迭代的后期采用停滞混沌变异避免算法陷入局部最优,并在算法的整个迭代过程中对算法参数进行动态自适应.利用5个Benchmark函数对改进算法的性能进行了测试,并与已存在的和声搜索算法进行比较,结果表明了改进算法的有效性.

The basic Harmony Search (HS) algorithm has good global search per-formance, but it also has the disadvantage of slow local convergence speed specially when the iteration solution approaches to the optimal solution. Considering this disadvantage, an Accelerated Harmony Search (AHS) algorithm is proposed. The idea of phase-optimization is proposed to improve the local convergence speed of the AHS algorithm. Different strategies are used during different iterative stages of the optimization algorithm. At the beginning, Chaos strategy is used to initialize the harmony memory. Cluster analysis strategy and chaos disturbance strategy are used in the mid-to-late stage. Chaos variation is used to improve the global optimal ability when the algorithm reaches stagnation in the late stage. Finally, parameter adaptive strategy is used during the whole iterative stage. The validity of the AHS algorithm is tested y five Benchmark functions. The results indicate that the AHS algorithm has better local convergence speed compared with those aleady proposed HS algorithms.