Sensor Fault Detection of a Highly Nonlinear CSTR Plant Using an Unscented Kalman Filter

In this paper, we propose an unscented Kalman filter (UKF) algorithm, as an alternative to the Extended Kalman Filter (EKF) for nonlinear processes fault detection. Although effectiveness of the EKF has been widely recognized, the practical applications of EKFs are still very limited. This is due to the fact that the estimates of EKF are often biased. Unscented filter is a new generalization of the Kalman filter for state estimation of nonlinear systems. In order to evaluate its ability, the presented method is applied to a highly nonlinear dynamic system describing the behavior of a non-adiabatic CSTR. The faulty behavior of output sensors in the chemical reactor is investigated. Simulation results show the efficiency and performance of the presented method.

Sensor Fault Detection of a Highly Nonlinear CSTR Plant Using an Unscented Kalman Filter Keywords:

Nonlinear dynamics , Simulation , Nonlinear state estimation , Extended Kalman filter , Unscented filter