基于引力搜索和粒子群混合优化算法的T-S模型辨识  

T-S Model Identification Based on a New Hybrid GSAPSO Algorithm

作　　者：唐柱[1] 丁学明[1] 刘灿[1] 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《上海理工大学学报》2013年第4期351-354,372,共5页Journal of University of Shanghai For Science and Technology

基　　金：国家自然科学基金资助项目(61074016)

摘　　要：提出了基于引力搜索(GSA)和粒子群(PSO)混合优化算法(GSAPSO)的T-S模型全局优化辨识方法.该方法充分整合GSA的勘探能力和PSO的开采能力,在GSA中引入PSO的个体最优值和群体最优值,同时改进惯性权重调整算法.T-S模型辨识分为结构辨识和参数辨识,采用聚类方法和GSAPSO算法同时辨识模型的结构和参数,从而实现全局优化辨识.仿真实例和比较分析证明了GSAPSO较标准的PSO和GSA有更强的全局优化能力和更高的辨识精度.A new hybrid algorithm GSAPSO based on the combination of particle swarm optimization （PSO）and gravitational search algorithm （GSA） was proposed for T-S model identification. The main idea is to integrate the ability of exploration of PSO with the ability of exploitation of GSA to synthesize the advantage of both algorithms： The PSO algorithm was introduced into GSA and an improved weight algorithm was presented. The structure identification and parameter identification of T-S model was realized together by using the new hybird algorithm GSAPSO and the clustering method. The results show the hybrid algorithm GSAPSO is of better capability of global optimization and higher precision than the standard PSO and GSA.

关 键 词：T—S模型 引力搜索算法 粒子群优化算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]