Online Model-Based Fault Detection and Diagnosis for a Smart Aircraft Actuator

Abstract This paper presents a real time diagnosis for an electrohydraulic aircraft actuator based on a nonlinear parameter estimation covering both system and sensor faults to facilitate a conditional maintenance smart actuator concept. The continuousdiscrete extended Kalman-Filter (EKF) is used with modifications and additions to the standard algorithm to achieve real time capability also for stiff differential equations, to make the parameter estimation robust against small or missing input and to fade out sensor hard over faults. The diagnosis can directly be performed from the estimated physical parameters. No supervision level for the parameter estimation or a extensive diagnosis decision is required. The algorithm is implemented on a DSP system and experimental results are obtained from an actuator test rig with artificially introduced faults.