基于一型模糊规则自主构建二型TSK神经模糊系统方法设计  

Automatically constructing type–2 TSK neural fuzzy system based on type–1 fuzzy rules

作　　者：高俊龙[1,2] 袁如意[1] 易建强[1,2] 应浩[3] 李成栋[4] GAO Jun-long;YUAN Ru-yi;YI Jian-qiang;YING Hao;LI Cheng-dong(Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China;University of Chinese Academy of Sciences, Beijing 100049, China;Department of Electrical & Computer Engineering, Wayne State University, Detroit 48101, USA;School of Information & Electrical Engineering, Shandong Jianzhu University, Jinan Shandong 250101, China)

机构地区：[1]中国科学院自动化研究所,北京100190 [2]中国科学院大学,北京100049 [3]美国韦恩州立大学电气与计算机工程学院,美国底特律48101 [4]山东建筑大学信息与电气工程学院,山东济南250101

出　　处：《控制理论与应用》2016年第12期1614-1629,共16页Control Theory & Applications

基　　金：国家自然科学基金项目(61421004;61403381;61473176);山东省属高校优秀青年人才联合基金项目(ZR2015JL021)资助~~

摘　　要：本文提出了一种使用一型模糊规则生成区间二型TSK(Takagi-Sugeno-Kang)神经模糊系统的新方法.该方法以训练数据集与使用自组织方法由该训练集训练生成的一型模糊系统为驱动,通过新型模糊系统前件类型转换算法与规则参数自适应学习算法的训练,在不高于原一型系统模糊集合总数前提下,自主构建区间二型TSK神经模糊系统.此外,针对两种典型的多输入单输出和多输入多输出系统,在3种不同强度的系统扰动场景下进行了对比仿真实验.实验结果表明,在含有不同扰动状态系统的建模与辨识中本方法较于对比方法具有更加优异的性能.This paper presents a novel approach to generating an interval type–2 TSK (Takagi-Sugeno-Kang) neural fuzzy system (IT2--TSK–NFS) by using type-1 TSK fuzzy (T1--TSK) rules. This method makes full use of training data sets and those T1 fuzzy rules generated from existing well-behaved self-organizing T1 methods to automatically generate a better performing IT2--TSK--NFS through novel antecedent type transformation and adaptive parameter training algorithms.Meanwhile, the rule number of the IT2--TSK--NFS stays the same as the original T1’s whereas the total number of IT2--FSs in the antecedent is no more than that of the original ones. Two benchmark examples with three different disturbance scenarios are given in experiments. The comparison results show and validate the proposed IT2--TSK--NFS can perform better than original T1--TSK system, and in some cases better than other IT2 self-organizing methods in literature in dealing with system modelling and identification issues under different disturbances.

关 键 词：二型模糊系统 神经模糊系统 类型转换 数据驱动 融合 

分 类 号：O159[理学—数学]