A multi-feature extraction technique based on principal component analysis for nonlinear dynamic process monitoring. (January 2020)

• Record Type: Journal Article
• Authors: Guo, Lingling; Wu, Ping; Lou, Siwei; Gao, Jinfeng; Liu, Yichao
• Is Part Of: Journal of process control. Volume 85(2020)
• Page Start: 159
• Page End: 172
• Publication Date: 2020-01
• Subjects: Principal component analysis -- Multi-feature extraction -- Nonlinear dynamic process -- Process monitoring; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2019.11.010
• Languages: English
• ISSNs: 0959-1524

Abstract

Highlights: The main contributions of the proposed MFPCA method are as follows: It takes the unique advantage of DiPCA, PCA and KPCA methods in modeling different features and then develops a monitoring index through combining these features for real-time process monitoring. The dynamic, linear and nonlinear features contained in the process data are considered simultaneously and the abnormal variations of these features can be reflected in the corresponding subspaces. Finally, the proposed MFPCA method can provide a better monitoring performance for nonlinear dynamic processes and it is easier to interpret the fault detection result. Abstract: Principal component analysis (PCA) and its modified methods have been widely applied in industrial process monitoring. In practice, industrial processes are with disparate characteristics, the process monitoring system should consider as many process characteristics as possible, such as dynamic and nonlinear characteristics. In this paper, a multi-feature extraction technique based on PCA is proposed for nonlinear dynamic process monitoring. The proposed method integrates dynamic inner PCA (DiPCA), PCA and kernel PCA (KPCA) methods through a serial structure to extract the dynamic, linear and nonlinear features among the process data. Along with the proposed method, the original data space is decomposed into several orthogonal subspaces, in which abnormal variations of different features can be monitored. For real-time process monitoring, a combined Hotelling's T 2 statistic based on the extracted multi-feature and a squared prediction error (SPE or Q ) statistic are established. Case studies on a numerical example and the Tennessee Eastman process are carried out to demonstrate the superior process monitoring performance of the proposed method compared with other relevant methods.