Research Of The Typical Fault Detection And Diagnosis Methods Of Winged Missile

In modern wars, the winged missile is increasing the performance of the control system and the stability of the variou parts more and more to enhance the combat capabilities under high-tech conditions. However, the actuators,sensors and so on in the missile flying may be out of work or lose some functions,such as the actuator seizure ,the sensor failure. So it is more important to detect the faults and find the fault positions timely and exactly. Thus the fault detection and diagnosis is playing the important significant roles in guaranteeing the missile survivabilities. Therefore it is also the key technology to keep the missile safety. In this paper, the methods of the fault detection and diagnosis are researched based on flight dynamics model of the winged model. The discussion is focusing on the some typical faults of actuators and sensors.Firstly, winged missile space motion mathematical model has been studied, simplified and linearized. Typical failures of the winged missile during flight have been analyzed. Follow in succession the RBF neural network(NN) is introduced which includes the amendations of own variations and the training modes.Sencodly, the adaptive observer is designed according to the linear state equation model. The design method is introduced and its stability is proved by Liapunov'theorem of stability. The method is proved usable to actuators and sensors faults through simulation according to the established linear missile sideway movement model.Thirdly, the RBF neural network observer is applied to diagnostic the failure based on the nonlinear missile model. In the first part, the off-line NN is made used to approximate the the model's nonlinear terms. The preliminary observer model is set up using the training result. The second part is building the on-line NN to approximate the fault values. The double parts consist of the RBF NN observer which is used to fault detection and diagnosis. In addition, a method is designed to gauge malfunction and judge the fault types. The simulations proves the designed methods workable.Finally, the winged missile fault detection and diagnosis simulation system is established combining with the nonlinear flight mechanics equations of a model missile. The system contains the model, PID controller, actuators saturation model and the state feedback. The RBF NN observer is proved practicable in fault detection and diagnosis for actuators and sensors. In the last of this section, the fault simulation interfaces is established using Matlab/GUI toolbox.