基于T-S模型的模糊系统辨识方法综述  

Overview of methods of fuzzy system identification on T-S model

作　　者：蒋强[1,2] 肖建[1] 何都益[2] 蒋伟[1] 王梦玲[1] 

机构地区：[1]西南交通大学电气学院,成都610031 [2]乐山师范学院,四川乐山614000

出　　处：《计算机应用研究》2009年第6期2008-2012,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(60674057)

摘　　要：模糊模型设计方法归结为两种,即语义驱动和数据驱动。数据驱动模型具有更好的性能,是目前研究的热点。模糊系统辨识是数据驱动下模糊系统建模的重要手段,辨识的优良直接影响系统建模的精度。模糊系统辨识可以分为两部分进行认识,即模糊系统结构辨识和参数辨识。回顾了近年来模糊系统辨识的理论和方法,如subtractive聚类、多分辨率自适应空间分解、SVM、核函数法、粒子群算法和并行遗传算法等。对各种算法原理、特点进行了介绍,对模糊系统辨识的发展进行了展望。There are different methodologies of fuzzy model design, which can mainly be divided into two groups ： the semantic-driven modeling and data-driven modeling. Data-driven modeling became more popular for better property than semanticdriven. Fuzzy system identification is one of the main approaches of fuzzy system modeling. The accuracy of fuzzy system model relate to the result of fuzzy system identification. Fuzzy system identification includes structure identification and parameters identification. This paper reviewed the state of theorem and methods of fuzzy system identification briefly. Disscussod widely used methods of fuzzy system identification, including those based on subtractive .clustering, multi-resolution analysis, support vector machine, kernel function, particle swarm and parallel genetic algorithm etc. Analyzed the characteristics of the methods, and outlined future research directions of fuzzy system identification.

关 键 词：模糊系统 系统辨识 结构辨识 参数辨识 T—S模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]