Adaptive neuro-fussy based control surface fault detection and reconfiguration

摘要

The aircraft becomes unstable due to fault in actuator or if there is a loss of control surface13; effectiveness due to damaged or blown surfaces. One of the popular methods to detect and reconfigure13; the surface fault is model based approach e.g. Extended Kalman filter (EKF). Using EKF, the parameters13; of control distribution matrix are estimated as augmented states of the system which are subsequently13; used to compute feedback gain to reconfigure the impaired system In this paper, detection and13; reconfiguration of surface fault in elevator of an aircraft is demonstrated using Adaptive Neuro - Fuzzy13; Inference System ANFIS . Under this approach, i) ANFIS is trained using time history of i/o data i.e.13; inputs as errors between nominal (healthy) states of aircraft and its faulty states (noise free) for different13; fault conditions and output as parameters of control distribution matrix and ii) trained ANFIS is13; subsequently used to estimate the parameters of control distribution matrix for the actual fault condition13; and the reconfiguration is carried out by computing new feedback gain using pseudo-inverse technique.