Abstract: In order to solve the premature convergence problem of Quantum-behaved Particle Swarm Optimization（QPSO）, a logistics Chaotic Mutation Quantum-behaved Particle Swarm Optimization（CMQPSO） is presented. Particles in population are first initialized using segmental Logistics chaotic mapping, and then particles are divided into two sub population-top population and bottom population based on their fitness values. Particles in top population are scattered with Gaussian disturbance when particles accumulate to a certain degree. Particles in bottom population are chosen by mutation probability and mutated with Logistics chaotic mapping, which in return, improve diversity of particles. Algorithm’s local and global search performance are well balanced with the introduction of mutation mapping and division of population. Results on Benchmark functions show that the proposed algorithm shows better search and convergence performance than standard QPSO and other algorithms. Effects of stagnation limit [Cσ] and proportion coefficient [S] on algorithm’s performance are analyzed in detail. And rational scope of the parameters is determined.

Key words: Quantum-behaved Particle Swarm Optimization（QPSO）, segmental Logistics mapping, mutation

摘要： 针对量子粒子群优化算法存在早熟收敛的问题，提出一种基于Logistics混沌映射变异的多种群量子粒子群优化算法（CMQPSO），采用分段Logistics混沌映射生成初始粒子群，根据适应度值将群体分为顶层和底层种群。顶层出现聚集时才进行高斯扰动，底层种群则按概率通过Logistics混沌变异生成分布更为均匀的粒子，提高种群的多样性，从而较好地平衡了算法的局部和全局搜索能力。对测试函数的计算表明算法较QPSO等其他算法在搜索能力和收敛速度方面有明显改进。分析了算法重要参数停滞阈值[Cσ]和比例系数[S]对搜索性能的影响，给出合理的取值范围。

关键词: 量子行为粒子群优化, 分段Logistics映射, 变异