Robust fault detection via unknown input observer based on cubature Kalman filter

In this paper, a novel procedure is proposed to design a nonlinear unknown input observer (NUIO) for robust fault detection purposes. NUIO decouples disturbances and uncertainties from the state estimation in nonlinear systems. The proposed method is based on cubature rules which are utilized in Cubature Kalman filter (CKF). The highlight of this work is to combine CKF with UIO to achieve a new NUIO structure for robust fault detection. For illustrating the performance of this approach, it is applied to a nonlinear continuous stirred tank reactor (CSTR) under faulty sensors. Compared to the previous NUIOs, which use extended Kalman filter (EKF) and unscented Kalman filter (UKF), simulation results demonstrate the superiority of the proposed approach.