一种双态免疫微粒群算法

A novel binary-state immune particle swarm optimization algorithm

针对基本微粒群算法的缺陷,提出了一种双态免疫微粒群算法.把微粒群分为捕食与探索两种状态,处于捕食状态的精英粒子采用精英学习策略,指导精英粒子逃离局部极值;处于探索状态的微粒采用探索策略,扩大解的搜索空间,抑制早熟停滞现象.同时引入免疫系统的克隆选择和受体编辑机制,增强群体逃离局部极值及多模优化问题全局寻优能力.实验表明该算法收敛速度快,求解精度高,尤其适合高维及多模态优化问题的求解.

Conventional algorithms of particle swarm optimization（PSO） are often trapped in local optima in global optimization. A novel binary-state immune particle swarm optimization algorithm（BIPSO） is proposed. In order to enhance the explorative capacity of the algorithm while avoiding the premature stagnation behavior, the meta-heuristics allow for two behavior states of the particles including Gather State and Explore State during the search. The population is divided into two parts in iterations. Elitist learning strategy is applied to the elitist particle to help the jump out of local optimal regions when the search is identified to be in a gather state. This paper propose a concept of explore strategy to encourage particle in a explore state to escape from the local territory. They exhibit a wide range exploration. Moreover, in order to increase the diversity of the population and improve the search capabilities of PSO algorithm, the mechanism of clonal selection and the mechanism of receptor edition are introduced into this algorithm. Experiments on several benchmarks show that the proposed method is capable of improving the search performance. It is efficient in tackling the high dimensional multimodal optimization problems.