摘要： 针对多维函数优化问题，提出2种新的反向自适应和声搜索算法。在自适应和声搜索算法的基础上，通过引入反向初始化操作，增强初始和声库的质量。设计一种反向自适应新和声搜索产生策略，加强算法的寻优能力。分别设计2种不同的和声微调概率的设置方式，并研究其对于算法收敛速度的影响。针对4个标准测试函数的仿真实验结果表明，与传统的和声搜索算法相比，2种算法的求解质量和收敛速度都有所提升。

关键词: 多维函数优化, 和声搜索算法, 进化计算, 连续优化, 反向学习, 自适应

Abstract: Two novel Opposition Self-adaptive Harmony Search(OSHS) algorithms are proposed to solve multi-dimensional function optimization problems. OSHS is based on Self-adaptive Harmony Search(SHS) algorithm. In order to enhance the quality of the initialized harmony memory, the opposition initialization is introduced in OSHS. An opposition self-adaptive mechanism is designed to enhance the exploration of OSHS. Two different ways are designed for setting the parameter. Meanwhile, the two ways’ impact on the convergence speed of OSHS is discussed. The simulation results for four benchmarks show that compared with the traditional search algorithms, the two new algorithms performe better in solution quality and convergence rate.

Key words: multi-dimensional function optimization, harmony search algorithm, evolutionary computation, continual optimization, opposition learning, self-adaptive

中图分类号: 

• TP301.6