Fault Diagnosis And Simulation Of Satellite Attitude Control System Based On Neural Network

With the increasing number of satellite launching missions of large Aerospace powers and private enterprises,new technologies and methods are emerging in endlessly,which makes the types of failures encountered in their on-orbit operation increase year by year.In order to realize the vision of mankind going out of the earth,there is an urgent need for advanced satellites with intelligent fault diagnosis system to diagnose the failure problems of satellites,so as to adapt to the complex space environment.In this paper,the satellite attitude control system is taken as the research object,and the fault diagnosis of actuators and sensors is studied.Based on the recurrent neural network,the corresponding fault diagnosis method is proposed,and the simulation platform of single-axis attitude fault diagnosis based on Aduino is designed.The main contents of this paper include:Aiming at actuator fault in satellite attitude control system,a fault diagnosis method based on recurrent neural network is proposed.The typical actuator,reaction flywheel,is modeled and analyzed,and two common faults are discussed.A recursive neural network parameter estimator is designed and the weight training algorithm is optimized.Thus,the multiple fault parameters of the flywheel are estimated,and the time and type of the fault are judged.Considering the coupling of fault sources,the diagnosis structure with feedback is further designed to improve the diagnosis accuracy.Aiming at the problem of sensor fault in satellite attitude control system,a fault diagnosis method based on recurrent neural network observer is proposed.Firstly,the common sensor-magnetometer is modeled and analyzed.For this purpose,a recurrent neural network observer and a learning rule of neural network adapted to sensor noise are designed,and its stability is proved by introducing Lyapunov function.By using the good approximation ability of recurrent neural network to the non-linear system,the fault of the system sensor is estimated,so that the fault time and the fault situation can be known,and finally the fault diagnosis of the sensor can be realized.Aiming at the fault diagnosis method proposed in this paper,a simulation platform for simulation verification is designed.In the aspect of hardware design,Arduino UNO is used as the main controller.At the same time,it is equipped with attitude measurement unit,Bluetooth communication unit and flywheel actuator.In the aspect of software design,the system model and attitude estimation of the simulation platform are given,and the control strategy of the platform is designed.Considering the measurement noise,the Unscented Kalman filter algorithm is used to reduce the noise.Through three sets of simulation experiments on the platform,the reliability and accuracy of the proposed fault diagnosis method are verified,and the effectiveness of the simulation platform is verified,which lays a foundation for further optimization and improvement of the diagnosis method.