Filtering of Linear Systems With Unknown Inputs

Abstract

State estimation of linear systems under the influence of both unknown deterministic inputs as well as Gaussian noise is considered. A Kalman like filter is developed which does not require the estimation of the unknown inputs as is customarily practiced. Therefore, the developed filter has reduced computational requirements. Comparative simulation results, under the influence of various types of unknown disturbance inputs, show the merits of the developed filter with respect to a conventional Kalman filter using disturbance estimation. It is found that the developed filter enjoys several practical advantages in terms of accuracy and fast tracking of the system states.