Fault Detection and Identification of Process Systems using Discriminant Analysis and Control Charts

摘要：

A new fault detection and identification methodology is designed and implemented for process systems. The methodology includes fault detection, fault identification and validation of fault. The two main technology used in this method are the Fisher discriminant analysis (FDA) and the XmR control chart. FDA reduces the original multi-dimensional variables to a single variable by maximizing Fisher criterion function. The transformed variables are monitored by XmR control charts. An application to the Tennessee Eastman process is illustrated.

关键词： Fault detection Fault identification Fault validation Fisher discriminant analysis XmR control charts

作者: Xudong Pei Yoshiyuki Yamashita Masatoshi Yoshida Shigeru Matsumoto

作者单位: School of Chemical Engineering and Environment,Beijing Institute of Technology,Beijing,100081,China Department of Chemical Engineering,Tokyo University of Agriculture and Technology,Tokyo 184-8588,Jap Department of Chemical Engineering,Tohoku University,Sendai 980-8579,Japan

会议类型: 国际会议

会议名称: 第四届亚洲过程系统工程会议暨2007年中国国际系统工程年会(The 4th International Symposium on Design,Operation & Control of Chemical Processes)(PSE ASIA 2007)

会议地点: 西安

会议语种:英文

在线出版日期: 2007-08-15（万方平台首次上网日期，不代表论文的发表时间）