一种基于数据驱动的模糊系统建模方法  

A Fuzzy System Modeling Method Based on Data-driven

作　　者：邵克勇[1] 范欣[1] 张永华[1] 田野[1] 

机构地区：[1]大庆石油学院电气信息工程学院,黑龙江大庆163318

出　　处：《化工自动化及仪表》2009年第5期25-28,共4页Control and Instruments in Chemical Industry

基　　金：黑龙江省博士后科研启动基金项目(LBH-Q08159)

摘　　要：针对工业生产中的一些复杂、非线性模糊系统,传统的建模方法很难描述其特性,而在实际生产中存在大量输入输出数据,提出了一种通用的基于数据驱动的模糊系统建模方法。采用减法聚类和模糊C-均值相结合的模糊聚类算法对输入空间进行划分,进而从输入输出采样数据中提取系统模糊规则,这样使得被辨识模型可用若干局部线性模型表示,然后利用递推最小二乘法对后件参数进行辨识,从而建立了非线性系统的T-S模糊模型。最后,应用该方法对一个非线性系统进行辨识,仿真结果验证了所提方法的有效性。For dynamic systems with complex and nonlinear characteristics,the traditional modeling methods couldn't describe its dynamic properties.In fact,there were a large number of input/output data in the complex systems,so a general fuzzy system modeling method based on data driven was developed.Combined subtractive clustering and fuzzy C-means clustering(FCM) to divide the input space,extracted fuzzy rules from the input and output data,then the identification model could be indicated by a number of local linear models,after that used recursive least squares method to identify the parameters,thereby a nonlinear system of T-S fuzzy model was set up.Finally,a nonlinear system was used to illustrate this method.Simulation results demonstrate the effectiveness of the proposed identification methods.

关 键 词：数据驱动 T-S模型 模糊C-均值聚类 递推最小二乘 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]