Investigation of Dynamic Multivariate Processes Monitoring
| Authors: | Xie Lei, National Key Laboratory of Industrial Control Technology, Zhejiang University, China
wang Shu-qing, National Key Laboratory of Industrial Control Technology, Zhejiang University, China
Zhang Jian-ming, National Key Laboratory of Industrial Control Technology,Zhejiang University, China |
|---|
| Topic: | 6.1 Chemical Process Control |
|---|
| Session: | Advances in Process Control |
|---|
| Keywords: | Multivariate statistical process control; Subspace identification, False alarms rate (FAR), Dynamic processes, Multivariate Statistical Process Control (MSPC) |
|---|
Abstract
Chemical process variables are always driven by random noise and disturbances. The closed-loop control yields process measurements that are auto & cross correlated. The influence of auto & cross correlations on statistical process control (SPC) is investigated in detail. It is revealed both auto and cross correlations among the variables will cause unexpected false alarms. Dynamic PCA and ARMA-PCA are demonstrated to be inefficient to remove the influences of auto & cross correlations. Subspace identification based PCA (SI-PCA) is proposed to improve the monitoring of dynamic processes. Through state space modelling, SI-PCA can remove the auto & cross correlations efficiently and avoid unexpected false alarms. The application in Tennessee Eastman challenge process illustrates the advantages of the proposed approach.
