Process Monitoring for Closed-Loop Control Systems With Incipient Faults

Incipient faults would evolve into permanent ones that lead to serious consequences without in-time detection and proper handling. Closed-loop control further complicates the task as it tends to compensate for incipient faults and further weaken their impact. Thus, in this article, a closed-loop process monitoring scheme for incipient faults is proposed. Inspired by the analysis of the fault impact in closed-loop process, correlation-based features for the closed-loop dynamic latent variable model is proposed for enhancing performance for detecting incipient faults. The features simultaneously shrink the nominal data variations and highlight the fault impact. A process monitoring strategy for incipient fault is, thus, proposed with the feature, and two case studies on the CSTR and the thruster system of the “Jiaolong” deep-sea submersible are carried on. The results show that the proposed method could detect incipient faults in closed-loop processes efficiently and prevent them from evolving to severe permanent faults, which helps ensure personnel safety and improve the maintainable level of the processes.