基于正交最小二乘估计的非线性系统模糊辨识  

Fuzzy Identification of Nonlinear System Based on Orthogonal Least Square

作　　者：王宏伟[1] 刘向利[1] 马广富[2] 

机构地区：[1]青岛海尔集团技术中心,青岛266101 [2]哈尔滨工业大学控制工程系,哈尔滨150001

出　　处：《计算机学报》2004年第8期1143-1146,共4页Chinese Journal of Computers

摘　　要：提出了一种基于正交最小二乘的模糊模型结构和参数辨识方法 .首先 ,基于正交最小二乘方法分析模糊模型的模糊关系矩阵 .通过分析正交向量在模型中贡献的大小 ,确定模糊模型的结构 ,即确定模糊模型的规则数、规则 .另外 ,再次通过正交最小二乘方法确定模糊模型的结论参数 ,实现模糊模型结构和参数的优化 .为了证明该方法的有效性 ,采用该文方法对Box Jenkins煤气炉数据系统进行建模研究 ,仿真结果表明该文方法能够对非线性系统进行辨识 .An identifying algorithm for the structure and parameters of fuzzy model is proposed based on an orthogonal least square. First of all, fuzzy relation matrix of fuzzy model is confirmed by orthogonal least square. The validity of fuzzy rules is made certain by means of analyzing the efforts of orthogonal vectors in fuzzy model, and then the number of fuzzy rules and fuzzy rules could be acquired. In addition, the conclusion parameters of fuzzy model are confirmed by the orthogonal least square in order to optimize the structure and the parameters of fuzzy model. To demonstrate the validity of the paper, the proposed method is successfully used to build Box-Jenkins's model of gas furnace.

关 键 词：正交最小二乘 模糊模型 模糊竞争学习 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]