Detection and estimation of sensor drifts using Kalman filters with a demonstration on a pressurizer

An algorithm for detection and estimation of sensor drifts is proposed in this paper. The algorithm is based on estimation of the process states from which the measurements are made and the rate of drifts using a state augmented Kalman filter. The detection and the estimation of a drift are carried out by evaluating the mean of the innovation sequence of the Kalman filter. The relationship between the mean and the drift is analyzed in detail to provide insights on the connection between the innovation sequence and the drift. The developed algorithm has been successfully applied to a pressurizer for detection and estimation of pressure sensor drifts. The results convincingly demonstrate the capability of the algorithm.

► How the expectation of the innovations changes in the drift case is formulated.
► Using the divergence in the expectation for detection of the drift is demonstrated.
► An augmented system model is proposed for estimation of the drift.
► Demonstration of the proposed algorithm is presented using a pressurizer model.