A novel method for actuator fault detection considering complex operating processes. (February 2022)

• Record Type: Journal Article
• Authors: Sun, Tianshu; Wang, Yinsong; Ding, Mengting
• Is Part Of: Journal of process control. Volume 110(2022)
• Page Start: 1
• Page End: 10
• Publication Date: 2022-02
• Subjects: Fault detection -- Actuator -- Steady and dynamic state -- Robust convex non-negative matrix factorization -- Kernel robust convex non-negative matrix factorization; Process control -- Periodicals; Fabrication -- Contrôle -- Périodiques; Process control; Periodicals; Electronic journals; 660.281
• DOI: 10.1016/j.jprocont.2021.12.003
• Languages: English
• ISSNs: 0959-1524

Abstract

Abstract: In the process industry, the actuator has complex operating processes. Its steady working conditions are easy to change. Moreover, it is in a dynamic process for long periods of time. These phenomena reduce the accuracy and timeliness of many condition monitoring (CM) methods. In this paper, a fault detection mechanism considering the complex operating processes of the actuator is proposed. One of its main feature is that it can detect faults under steady state regardless of variable load changes. The other is that it has ability to detect faults during dynamic processes refusing the interference of nonlinearity and dynamic characteristics. Firstly, an actuator data model library based on historical data is established offline. Secondly, in the online situation, after the R indicator divides the testing data into steady state or dynamic, it is input into the model library to obtain a standard training set. Finally, the steady-state/dynamic fault detection module will determine whether a fault has occurred based on the testing data and the training set. Moreover, open-loop experiments based on the DAMADICS (Development and Application of Methods for Actuator Diagnosis in Industrial Control Systems) platform and closed-loop experiments based on the water tank system are designed. These experiments show that, compared with other methods, the proposed fault detection framework has better monitoring performance during complex operating processes of actuators. Highlights: A novel actuator fault detection method is proposed. It can deal with various complex operating processes containing steady and dynamic state. It contains two novel extended non-negative matrix factorization algorithms. The presented mechanism is verified by simulations and experiments.