Robust probabilistic predictable feature analysis and its application for dynamic process monitoring. (April 2022)

• Record Type: Journal Article
• Authors: Fan, Wei; Zhu, Qinqin; Ren, Shaojun; Zhang, Liang; Si, Fengqi
• Is Part Of: Journal of process control. Volume 112(2022)
• Page Start: 21
• Page End: 35
• Publication Date: 2022-04
• Subjects: Robust dynamic process monitoring -- Probabilistic predictable feature analysis -- EM algorithm -- Genetic algorithm -- Kalman filter; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2022.02.004
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: Dynamic process monitoring with multivariate statistical methods has been widely researched and applied for anomaly detection in dynamic systems. However, most of them are designed under the assumption that the measurement noise follows a Gaussian distribution, which usually contradicts the actual situations. Alternatively, in this paper, a novel mixture of Gaussian and Student's t distributions is designed to account for the system noise, and a robust probabilistic predictable feature analysis (RPPFA) algorithm is proposed to capture both the process dynamics and the characteristic of the heavy tail from multivariate temporal data. Moreover, an improved Expectation–Maximization algorithm and a modified Kalman filter method are proposed for parameter optimization. Three monitoring statistics are designed to identify abnormal conditions, which are Hotelling's T 2, squared prediction error and dynamic index. The superiority of RPPFA is demonstrated through two case studies, namely, a simulated three phase flow facility and an industrial medium speed coal mill. Highlights: A RPPFA is proposed to model temporal series with the existence of outliers. Dynamics are captured by multi-step ahead prediction of latent variables. Modified Kalman filter is proposed to evaluate latent variables' expectations. The robustness of RPPFA is verified through two industrial applications.