基于K-均值聚类的免疫粒子群优化算法  

Immune Particle Swarm Optimization Algorithm based on K-means Clustering

作　　者：满春涛[1] 张锐[1,2] 张涛[3] 

机构地区：[1]哈尔滨理工大学自动化学院,黑龙江哈尔滨150080 [2]哈尔滨工业大学电气工程及自动化学院,黑龙江哈尔滨150010 [3]哈尔滨理工大学电气与电子工程学院,黑龙江哈尔滨150040

出　　处：《控制工程》2010年第6期796-799,共4页Control Engineering of China

基　　金：黑龙江省自然科学基金资助项目(F2007-9);哈尔滨市科技创新人才研究专项资金项目(2008RFXXG009);黑龙江省教育厅科研项目(11531085)

摘　　要：在解决复杂函数优化问题时,免疫粒子群优化算法在疫苗提取操作中,通常把最佳粒子作为有效特征信息,没有从根本上解决"趋同性"问题,影响算法的收敛速度和收敛精度。针对该问题,提出了一种基于K-均值聚类的免疫粒子群优化算法(KIPSO),将K-均值聚类算法用于免疫疫苗的提取,确定最高平均适应值的聚类中心及其最大邻域,得到符合具有最优个体特征的疫苗集,并以自适应方式确定疫苗集进行疫苗更新,提高算法的收敛性。仿真结果表明,与免疫粒子群优化算法相比,该算法提高了收敛速度和收敛精度。To the problem that,in complex function optimization,the optimal particle is regarded as effective information in process of vaccine obtaining for immune particle swarm optimization algorithm,and cannot solve the problem of ＂tendency＂,influencing convergence speed,and convergence precision,a K-means immune PSO is proposed,which is used for extracting vaccine,identifying clustering center and its maximum neighborhood.The most superior individual characteristic vaccine set is obtained,and the extracting vaccine is updated by adaptive method.The algorithm is improved in convergence.The simulation results show that the proposed method enhance convergence speed and convergence precision comparing with immune particle swarm optimization algorithm.

关 键 词：K-均值 疫苗提取 免疫粒子群优化算法 自适应 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]