局部自回归模糊模型及其在小样本数据系统建模中的应用  

A Local Auto-Regressive Fuzzy Model and Its Application in Modeling of Small Samples Data Systems

作　　者：张洪[1] 萧德云[1] 刘震涛[2] 

机构地区：[1]清华大学自动化系,北京100084 [2]清华大学台湾研究所,北京100084

出　　处：《系统工程》2002年第4期91-96,共6页Systems Engineering

基　　金：国家自然科学基金资助项目 (70 0 4 10 0 3)

摘　　要：提出一种基于 Takagi- Sugeno模型 (TS模型 )结构的局部自回归模糊模型 (L ARF模型 ) ,将其用于小样本数据情况下的系统建模。深入分析 LARF模型的结构特征、前件的选取、后件参数的辨识及其评价指标等 ,并提出了完整的模型辨识算法。该建模方法从一维的角度来考虑输入输出空间 ,减少待辨识参数的个数 ,同时采用一种具有两个调节参数的隶属函数 ,在求隶属度的同时完成了数据处理 ,从而可以省略数据预处理的工作。实例研究表明 ,LARF模型是一种适合用来描述小样本数据系统的模型 。A local auto regressive fuzzy model (LARF model) based on Takagi Sugeno model (TS model) structure is put forward in this paper. Its application in the modeling of small samples data systems is analyzed. The structure of the model, determination of the premise, identification of the consequence and the criterion of the LARF model are discussed in detail. The detailed identification algorithm is also brought forward. The input output universe can be considered as one dimension,so the number of the parameters needed to be identified is decreased and the identification algorithm is also simple. Moreover, a suitable membership function with two tuning parameters is obtained, which can process the original data while calculating their matching degrees and the pretreatment is unnecessary. The results of the case show that the LARF model is suitable for the small sample data systems and both the modeling method and the identification algorithm are practical.

关 键 词：局部自回归模糊模型 小样本数据系统 建模 TS-模型 LARF模型 系统工程 

分 类 号：N945[自然科学总论—系统科学]