Fault Diagnosis and Accommodation of Linear Stochastic Systems with Unknown Disturbances

Jonghee Lee,J. Lyou

Abstract:An integrated robust fault diagnosis and fault accommodation strategy for a class of linear stochastic systems subjected to unknown disturbances is presented under the assumption that only a single fault may occur at a given time. The strategy is based on the fault isolation and estimation using a bank of robust two-stage Kalman filters and introduction of the additive compensation input for cancelling out the fault’s effect on the system. Each filter is set up such that the residual is decoupled from unknown disturbances and fault with the influence vector designed in the filter. Simulation results for the simplified longitudinal flight control system with parameter uncertainties, process and sensor noises demonstrate the effectiveness of the present approach.