In order to improve the reliability

In order to improve the reliability and fault tolerance of aileron actuator fault detection, a fault detection method for aileron actuator using RBF neural network is proposed. In the variable conditions, the inputs of systems, the outputs of systems, FPGA Mezzanine Card (FMC), aerodynamic load detecting faults and three RBF neural networks are used. The first neural network is used as an observer to estimate aileron actuator outputs; the second neural network is used to adjust the adaptive threshold according to the processing conditions. Finally, the third neural network is used as a FMC observer to output the estimation FMC. The mechanical failure is detected from the comparison of the estimated FMC and the actual FMC. The advantage is that the adaptive threshold and the residuals can be effectively coordinated, and adaptive to variable conditions. Simulation results show that the proposed method can detect aileron actuator failures well. When a mechanical fault is introduced into the system, the estimated FMC deviates the actual FMC. When an electrical fault is introduced into the system, the estimated FMC approximately equals to the actual one. Key words: Aileron actuator; fault diagnose; RBF neural network; FMC observer; electrical fault; mechanical fault 1