面向多峰函数的自适应小生境量子进化算法

Adaptive niche quantum evolutionary algorithm for multimodal function

为解决量子进化算法在多峰优化时只能找到一个最优解,无法找到所有全局和局部最优解的问题,提出自适应小生境量子进化算法。利用佳点集理论初始化种群,使种群均匀分布在整个搜索空间;提出中心地形信息小生境自适应识别方法,用于自适应的识别峰值所在区域,并建立小生境完善策略,提高小生境识别速度;借助量子进化算法的快速寻优能力精确寻找各个峰值点;采用动态种群调整策略,维持种群的多样性,自适应地调节种群规模。仿真实验结果表明,该算法具有较强全局优化能力和局部优化能力,且搜索到的每个最优解都达到了理想值。

Since it is difficult to find all the global and local optimal solutions in multimodal optimization problem for quantum evolutionary algorithm which can only find a global optimal solution, an adaptive niche quantum evolutionary algorithm is proposed. A good-point set is used to produce the initial population which is scattered uniformly over the entire search space. An adaptive niche identification method based on topographic center is designed to identify the extremum areas of the population adaptively, and a strategy of niche integrity is presented to increase the niche identification speed. The fast optimization ability of quantum evolutionary al- gorithm is applied to search extrema precisely. The strategy of dynamic population has been used to maintain diversity of population, and adjust the size of population adaptively. Simulation results show that the proposed algorithm has good glabal optimization performance and local extremum search ability and solutions are satisfactory.