基于灰聚类多子群自适应PSO算法的Volterra核辨识  

Identification of Volterra series based on grey clustering multi-subpopulation adaptive PSO algorithm

作　　者：李宁洲[1,2] 冯晓云[1] 

机构地区：[1]西南交通大学电气工程学院,成都610031 [2]兰州交通大学机电工程学院,兰州730070

出　　处：《计算机应用研究》2014年第6期1697-1701,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(51277153;11162007);甘肃省自然基金资助项目(1308RJZA149)

摘　　要：针对一类单输入单输出非线性动态系统的Volterra级数模型辨识问题,提出了灰聚类多子群自适应PSO算法,并定义了精度影响系数以定量评估模型结构项对辨识精度的影响程度。在利用Volterra级数对非线性系统进行初始建模的基础上,采用灰聚类多子群自适应PSO算法和精度影响系数实现了非线性Volterra级数模型的结构确认和参数优化辨识。将该方法与基于标准PSO、GA、QPSO算法的Volterra时域核辨识方法进行了对比实验。结果表明,该方法在辨识精度及收敛速度等方面明显优于其他方法。For Volterra model identification of a single-input and single-output nonlinear dynamic system,this paper put forward the grey clustering multi-subpopulation adaptive PSO algorithm and defined accuracy influence coefficient to assess quantitatively the degree of influence that the model structure items impact on the accuracy of identification.At first,the method used Volterra series to create the initial model of nonlinear system,and then,it adopted grey clustering multi-subpopulation adaptive PSO algorithm and accuracy influence coefficient to identify the model structure and parameters.The simulation results show that the proposed method in such aspects as identification precision and convergence speed is superior to other Volterra series identification methods based on standard PSO algorithm,GA and QPSO algorithm methods.

关 键 词：灰聚类多子群自适应PSO算法 VOLTERRA级数 精度影响系数 非线性系统辨识 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]