基于自适应变异的动态粒子群优化算法  

Dynamic Particle Swarm Optimization Algorithm Based on Adaptive Mutation

作　　者：王改堂[1] 李平[2] 苏成利[2] 

机构地区：[1]西北工业大学自动化学院,西安710072 [2]辽宁石油化工大学信息与控制工程学院,辽宁抚顺113001

出　　处：《科技通报》2010年第5期657-660,665,共5页Bulletin of Science and Technology

基　　金：国家高技术研究发展计划(863计划)项目(2007AA04Z162);辽宁省高等学校优秀人才支持计划资助(2008RC32)

摘　　要：提出了一种新的基于自适应变异的动态粒子群优化算法。该算法除了采用动态惯性权重外,还引入了自适应学习因子和新的变异算子。该算法在运行过程中,根据群体适应度方差以及当前最优解的大小来确定当前最佳粒子的变异概率,采用新的变异算子变异增强了该算法跳出局部最优解的能力。对几种典型函数的测试结果表明:新算法具有很强的全局搜索能力,收敛速度和收敛精度也有所提高,并且能有效避免早熟收敛问题。A new dynamic particle swarm optimization algorithm based on adaptive mutation is presented. Besides that it makes use of adaptive inertia weight, adaptive term coefficients and new mutation operator were introduced to the algorithm. During the running time, the mutation probability for the current best particle is determined by two factors： the variance of the population＇s fitness and the current optimal solution. This paper makes use of new mutation operator for mutation, the ability of the algorithm to break away from the local optimum is greatly improved. The proposed algorithm is tested with four well-known benchmark functions. The experimental results show that the new algorithm has great global search ability, convergence accuracy and convergence velocity is also increased, and avoid the premature convergence problem effectively.

关 键 词：粒子群 惯性权重 自适应 变异算子 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]