摘要
针对经典粒子群(PSO)算法易出现早熟收敛和搜索精度差的缺陷,提出了一种基于混沌变异的k-均值聚类PSO优化算法(FCPSO)。该算法首先通过k-均值聚类方法把粒子群分成若干个子群体,从而在迭代过程中每个粒子根据其个体极值和所在子种群中的全局极值来更新自己的位置和速度。其次,在算法中引入自适应混沌变异,有效的增强了子群体之间信息交换和经典PSO算法跳出局部最优解的能力。对几个典型可变维函数的测试结果表明,该算法是非常有效的。
A new k-mean cluster particle swarm algorithm （FCPSO） based on self-adaptive chaotic mutatiori is presented to overcome the default of the premature and low precision of the standard PSO algorithm. First, the particle swarm is divided into several sub-swarms by the k-mean cluster. Then, the current particles are dynamically updated by the personal best particle and global best particles in the sub-swarms. Second, by the self-adaptive chaotic mutation operator introduced to the algorithm, the information exchanged between different sub-swarms and the ability of standard PSO algorithm is break away from the local optimum are effectively improved. The computer simulations demonstrate the effectiveness of the proposed algorithm.
出处
《科学技术与工程》
2009年第5期1150-1154,共5页
Science Technology and Engineering
基金
陕西省自然科学基础研究计划项目(2006A12)
宝鸡文理学院重点科研项目(ZK0619)资助
