模糊聚类与最小二乘相结合建立非线性系统模型

MODELING NONLINEAR SYSTEMS USING FUZZY CLUSTERING ASSOCIATING WITH RLS

提出—种模糊聚类与最小二乘相结合的辨识方法。该方法利用基于模糊似然函数的模糊聚类算法确定系统的模糊划分数目,进而对应聚类个数建立相应的Takagi-Sugeno局部线性化模型,并结合递推最小二乘法,完成系统的辨识。该方法可使模糊模型的结构辨识和参数辨识同时完成,从而实现模糊模型的在线辨识。该方法辨识速度快,精确度高。仿真结果验证了该方法的有效性。

In this paper, a new method of nonlinear system modeling using fuzzy clustering associating with RLS is presented. A nonlinear system can be quickly divided into several fuzzy parts using clustering algorithm based on the fuzzy likelihood function. Regarding cluster number as a rule number, Takagi-Sugeno fuzzy model can be built. Then the system is identified associating with RLS. The proposed method can accomplish the structure and the parameter identification of the fuzzy model in the same time, and implements the on-line identification of the fuzzy model. The simulation results prove the effectiveness of the method.