Research On Intelligent Fault Diagnosis Of Spacecraft Attitude Control System

The rapid development of space technology has changed people's way of life.But the environment in space is complex and changeable,making the spacecraft very prone to failure.The attitude control system 1s an important subsysten to ensure the stable operation of the spacecraft.The probability of actuators and sensors in the system failure is high.The application of fault diagnosis technology in attitude control systen can effectively respond to system faults,avoid more serious faults,and enable spacecraft to operate safely and steadily in space.Based on the intelligent method,this paper focuses on the fault diagnosis of actuators and sensors in spacecraft attitude control system.The main work and research contents of this paper include:(1)On the basis of consulting a large number of documents,the composition of spacecraft attitude control system is analyzed concretely.According to the parameter conversion relationship between each module of the spacecraft attitude control system,the simulation model of the spacecraft attitude control system is built by using MATLAB/Simulink tool.The simulation model can generate data for analyzing the attitude control system.The common faults of actuators and sensors are modeled separately.To lay the foundation for the following chapters of research work.(2)The intelligent method based on BP neural network is applied to fault diagnosis of actuator wheel in attitude control system.The application of BP neural network in fault diagnosis 1s analyzed.A wheel fault diagnosis model 1s established.The input of the model is the voltage acting on the wheel,and the output is the attitude data of the spacecraft.The model is simulated by MATLAB.The real attitude data of the spacecrah are compared with the attitude data predicted by the model to diagnose the wheel fault.The diagnosis results show that the wheel fault can be correctly diagnosed.(3)The intelligent method based on LSTM network is applied to fault diagnosis of gyroscope and star sensor in attitude control system sensors.The application of LSTM network in fault diagnosis is analyzed.The fault diagnosis models of gyroscope and star sensor are established respectively.The model predicts the sequence data generated by gyroscope and star sensor.The sensor fault is diagnosed by comparing the real output data of sensors in spacecraft with the predicted data of models.The diagnostic results show that sensor faults can be correctly diagnosed.(4)Based on the semi-physical simulation system platform of spacecraft and the actuator wheel,the BP neural network fault diagnosis model is tested and analyzed.The experimental results show that the wheel fault can be diagnosed.