Fault detection of Tennessee Eastman process based on topological features and SVM

2017 2nd International Conference on Mechatronics and Electrical Systems

机械制造;无线电电子学

Zhao, Huiyang^1,2 ; Hu, Yanzhu^1 ; Ai, Xinbo^1 ; Hu, Yu^1 ; Meng, Zhen^1

Beijing Key Laboratory of Work Safety Intelligent Monitoring, Beijing University of Posts and Telecommunications, Beijing

100876, China^1

School of Information Engineering, Xuchang University, Xuchang

461000, China^2

关键词: Complex network models;    Global informations;    Industrial processs;    Industrial systems;    Network construction;    Research topics;    Tennessee Eastman process;    Topological features;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/339/1/012039/pdf
DOI  :  10.1088/1757-899X/339/1/012039

来源: IOP

【 摘 要 】

Fault detection in industrial process is a popular research topic. Although the distributed control system(DCS) has been introduced to monitor the state of industrial process, it still cannot satisfy all the requirements for fault detection of all the industrial systems. In this paper, we proposed a novel method based on topological features and support vector machine(SVM), for fault detection of industrial process. The proposed method takes global information of measured variables into account by complex network model and predicts whether a system has generated some faults or not by SVM. The proposed method can be divided into four steps, i.e. network construction, network analysis, model training and model testing respectively. Finally, we apply the model to Tennessee Eastman process(TEP). The results show that this method works well and can be a useful supplement for fault detection of industrial process.