基于区间T-S模糊模型的自适应故障检测算法研究  

Adaptive Fault Detection Based on Interval T-S Fuzzy Model

作　　者：刘小雍[1] 熊中刚[1] 阎昌国[1] 罗光毅[1] 陈孝玉[1] LIU Xiaoyong XIONG Zhonggang YAN Changguo LUO Guangyi CHEN Xiaoyu(College of Engineering and Technology, Zunyi Normal College, Zunyi, Guizhou 563002, Chin)

机构地区：[1]遵义师范学院工学院,贵州遵义563002

出　　处：《新疆大学学报（自然科学版）》2017年第2期230-236,共7页Journal of Xinjiang University(Natural Science Edition)

基　　金：贵州省教育厅青年项目(黔教合KY字[2016]254;黔教合KY字[2015]457);遵义师范学院博士项目(遵师BS[2015]04);贵州省科学技术基金(黔科合LH字[2016]7003号;黔科合LH字[2015]7054号;黔科合LH字[2016]7018号;黔科合LH字[2016]7002号)

摘　　要：针对传统的故障检测方法通常事先人为设定一个健康状况下的确定阈值来实现故障检测,往往不具有故障的自适应检测能力,本文提出了基于区间T-S模糊模型的自适应故障检测方法,即系统在运行过程中的不同时刻,系统无故障状态对应的阈值是不一样的.该方法首先将故障检测的问题转化为基于逼近误差的l1范数最小化的区间模型优化问题,应用线性规划对由T-S模型构成的区间[f L(t),f U(t)]参数进行求解.所建立的区间模糊模型将用于判断故障是否发生.最后,通过Tennessee Eastman(TE)过程论证了提出方法对故障检测的自适应能力和实时性.Considering that the tradition fault detection method using the definite threshold controlled by some approaches is incapable of detecting fault adaptively,this paper proposes an adaptive fault detection approach based on interval T-S fuzzy model.Characteristics on adaptive are reflected by different threshold.Firstly,the proposed method translates the problem on fault detection into optimization with interval model,and uses l1-norm on Approximation Error to construct optimization corresponding to interval model and this parameters consisted of fL（t） and fU（t）based on T-S model are solved by linear programming.Interval T-S fuzzy model is to be applied to judge whether the fault is detected or not.Finally,adaptive ability and real-time performance of detecting fault are demonstrated by the proposed method in the Tennessee Eastman process.

关 键 词：区间T-S模糊模型 线性规划 故障检测 逼近误差的l1范数 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]