Abstract:

Aiming at the obvious aerodynamic parametric uncertainties of aircraft in the full flight envelope, a fault reconstruction method based on a modified adaptive observer is proposed. The adaptive observer is improved by the uncertain parameter estimation process and validated on the longitudinal model of some unmanned vehicle. According to this method, the longitudinal model is described as an affine nonlinear structure with time variant parameters, and the observability of its parameter-augmented model is analyzed. On this basis, to deal with conservatism of the robustness dead zone technique in fault detection and refine the detection sensitivity, aerodynamic parameters are identified on line by the augmented cubature Kalman filter (ACKF) algorithm. Then the parameter estimations are adopted in adaptive observer design. As the Lie derivative based criterion guarantees observability of the object system, the literature restriction that the object system must be within some particular structure is avoided. Based on this, the adaptive fault detection threshold and the fault parameter regulation law are derived, and the convergency of the estimation error is analyzed. Finally, simulations are conducted to testify the availability of this method.