Recursive Innovational Component Statistical Analysis for Fault Detection in Dynamic Processes

Abstract

This paper proposes a new data-driven process monitoring method named recursive innovational component statistical analysis (RICSA) for dynamic processes. RICSA divides the original data into dynamic components and innovational components, and then monitors the innovational components using the recently proposed recursive transformed component statistical analysis (RTCSA), which is effective for incipient fault detection. Compared to the dynamic version of RTCSA, recursive dynamic transformed component statistical analysis (RDTCSA), RICSA can deal with unsteady state, meanwhile, it has the lower calculation, the higher detection rate and the lower false alarm rate because of the reasonable space division. Through the experiment on the numerical simulation, the superiority of RICSA is verified.