针对标准粒子群算法在处理复杂函数时存在的收敛速度慢、易陷入局部最优的缺点,提出了新的混合粒子群算法.该算法利用混沌运动的遍历性、对初始条件的敏感性等特性进行群体的混沌初始化,且捕食搜索策略可以通过调节限制级别的控制粒子群的搜索空间,从而平衡全局搜索和局部搜索.测试结果表明,新算法具有更快的收敛速度和更强的全局寻优能力.
This paper proposed an effective hybrid particle swarm optimization merging chaos and predatory search. The new algorithm used the periodicity, and regularity of chaos to im- prove individual quality; predatory search can control the search space of the particle swarm though adjust the level of restriction. Thereby, the global search and local search can be bal- anced. Local search could make arithmetic convergence faster and more accurate. The test results showed that the new algorithm provided faster and stronger global optimization ability.