Dual redundancy fault diagnosis and reconstruction system of sensors based on BP neural network

Aircraft braking system is the key to ensure the safety of aircraft take-off and landing, and it is the final safety barrier of flying. In the aircraft brake system, pressure sensors are installed. However, the sensor itself is fragile and sensitive, which is prone to failure in the harsh flight environment. If the fault of the sensor itself is not handled well, giving the wrong indication may lead to serious consequences. As the input of the brake control system, the research of fault diagnosis and reconstruction technology for sensors is helpful to improve the reliability and safety of the control system. In this paper, a dual redundancy fault diagnosis and reconstruction system based on BP neural network is designed. The system can diagnose the fault of the sensor signal, reconstruct the fault sensor signal, and output the most appropriate fault free value to the subsequent control system to ensure the normal operation of the control system. The signal of pressure sensor in aircraft brake system is simulated and analyzed. The simulation results show that the designed network training error is basically less than 0.05Mpa (0.5%), and the local error is less than 0.15Mpa (1.5%). In the case of a paranoid failure of the pressure sensor, the decision-making module can realize the function of fault diagnosis and reconstruction, and output a fault-free signal, which proves the effectiveness of the method.