Abstract: In dynamic systems, the failure of a sensor or controller may lead to serious deterioration of system performance and even make the system not work normally. At present, a state observer or filter is mainly applied to the detection of sensor or controller failure to provide analytic redundancy. However, in real systems there are many uncertain factors. During the design of a state estimator, one can meet the problem that the state estimates deviate from the real values of the uncertainties of parameters are neglected.In this paper, the detection filter theory is applied to the design of the fault detector and the fault parameter identifier for the aircraft with the internal element or control element fault. Because Lyapunov's stability theory is applied, the fault tolerant control law can ensure the stability of the flight control system.A model reference fault tolerant control system is designed and simulated using a short period longitudinal model of a certain aircraft. Because the flight range of the aircraft is very large, its parameter variation is very violent. When the aircraft is normal, a parameter robust controller in the system is designed in order to obtain good performances under every flight state.

Key words: fault tolerance control, model reference, fault detector, parameter identifier, stability