State Estimation and Sensor Bias Detection Using Adaptive Linear Kalman Filter

For safe and economical process operations, state estimation of dynamic systems is an important prerequisite, since all the process variables are generally not measurable. Keeping this in view, a linear and non-linear observer has been developed to estimate the state of a noisy dynamic system. Also chemical processes have been suffering from several faults since many decades. Finding the system faults and isolating them is essential because if a fault occurs in the plant, there will be serious economic losses due to drop in productivity. To serve this purpose, an Adaptive Linear Kalman Filter (ALKF) and an Extended Kalman Filter (EKF) have been designed to detect and isolate the faults. Of all the faults, sensor faults are taken into consideration in this paper. Our task is to compare the results of two estimators and to identify the better one, in estimating the state of non-linear system along with bias detection. Simulation studies are also included to estimate the residuals of the estimators.