一种基于T-S模糊模型的残差加权故障诊断方法  

A Residual Weighting Fault Diagnosis Method Based on T-S Fuzzy Model

作　　者：杨马英[1] 张书桂 

机构地区：[1]浙江工业大学信息工程学院,浙江杭州210023

出　　处：《计算机与应用化学》2017年第10期802-808,共7页Computers and Applied Chemistry

摘　　要：为有效利用过程数据提升故障诊断准确性,提出基于系统T-S模糊模型的残差加权故障诊断方法。该方法首先采用T-S模糊模型建立系统正常工况下的解析模型,然后计算不同故障数据集在该模型下残差序列,以残差序列集合的中心为故障的基准残差序列,并以各故障数据集的标准差作为标示该故障域。最后通过模糊聚类以最大隶属度所在故障类型为最终故障类型快速诊断故障源。文章以TE过程为研究对象进行仿真,并选取其中4个故障类型进行故障诊断研究,仿真结果验证该诊断方法的有效性。A sort of weighted residual method based on T-S fuzzy model was proposed for fault diagnosis problem. Firstly, several multiple subsystems＇ analytic models are established under normal operating conditions using T-S fuzzy model. Secondly, the residual vector of different fault data sets in the model are calculated, with the residual sequence set＇s center as the standard fault residual sequence, and the faults coverage are marked referring to the fault data sets＇ standard deviation. Finally, by fuzzy C-means clustering, the fault type with maximum membership degree is rapidly diagnosed as the final fault source. The Tennessee Eastman benchmark process （TE） was examined with four fault types, and simulation results verified the diagnostic method＇s feasibility.

关 键 词：故障诊断 动态建模 参数识别 过程系统 T-S模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]