Fault detection and diagnosis using cubature Kalman filter for nonlinear process systems

Abstract

This paper presents a fault detection and diagnosis (FDD) for a nonlinear systems using multiple Cubature Kalman Filter (CKF) model. The proposed scheme able to identify sensors and actuators fault even with the presences of process and measurement noise. Comparison between actual faults with expected fault trajectory enables the FDD to narrow down possible scenario. The utilization of continuous stirred tank reactor (CSTR) simulation illustrates the performance of the scheme in nonlinear system. Result of the study shows the proposed method works effectively in determine the type of fault occurs in the CSTR.