摘要： 基于粒子群优化算法种群结构相对独立的特点，提出了一种改进的粒子群优化算法——随机摄动粒子群优化算法。该算法通过对每一次进化计算后记忆中的最优粒子进行随机摄动操作来提高解的精度和算法的搜索效率，同时通过对种群中的最差粒子重新进行初始化来保持种群的多样性以避免陷入局部最优解。通过典型复杂函数测试表明，随机摄动粒子群优化算法的优化性能和效率远远超过基本粒子群优化算法。

关键词: 粒子群优化算法；随机摄动；进化种群多样性

Abstract: A novel particle swarm optimization algorithm——random perturbation particle swarm optimization algorithm(RP-PSO) based on independence of population structure is proposed. To retain diversity of population and avoid being plunged to local optimum, it initializes the worst individual in population over again, at the same time, the best previous particle of each individual is randomly perturbed after evolutionary computation every time to improve its running efficiency and precision of over all optimization searching. Test results of complex functions demonstrate RAPSO is superior to basic particle swarm optimization in quality and efficiency.

Key words: Particle swarm optimization algorithm; Random perturbation; Diversity of evolution population