UIO-based diagnosis of aeroengine actuator faults

In order to improve the robustness of aircraft engine diagnosis against noisy disturbances and parameter variation, this paper addresses the problem of diagnosing actuator fault in aeroengine control systems. The proposed approach applies the unknown input observer (UIO) theory to estimate operational states in an operating aeroengine. The effect of exogenous disturbances on state estimation is mitigated in an orthogonal-projection-of-disturbances manner. As for engine actuators, we designed a group of UIOs, each of which is used to monitor the amount of faulty offset in its corresponding actuator. A residual vector between the estimated values and the measured outputs is further evaluated. In addition, the characteristics in the magnitude of the residuals are analyzed to achieve the detection and isolation of faults in aeroengine actuators. Experimental results on some turbofan engine indicate that the UIO-based method can more robustly detect and isolate actuator faults in an aeroengine than the Kalman filter-based algorithm.