BP Neural Networks Based Sensor Fault Diagnosis And Sliding Mode Fault-Tolerant Control Of Aero-Engines

Aero-engines are a kind of complicated nonlinear systems. In the whole flight envelope, the operating state of the engine changes greatly. Therefore, the research on sensors fault diagnosis and fault-tolerant control of aero-engine is very important, which can improve the safety and reliability of the engines control system. This dissertation describes the engine fault diagnosis and fault tolerant control problems on the basis of Back Propagation neural network and sliding mode control(SMC).This dissertation discrete double-input double-output engine state variable model is established based on the component-level model by the use of discrete small perturbation method to calculate the initial value of the linear model, fitting component-level model step data to obtain the engine discrete model. On the basis of aero-engine linear model, training samples are obtained to established estimation model based on BP neural network. Then desing the fault diagnosis system of engine sensors.PID reaching law based on the sliding mode controller was designed for study the effect of the sliding mode controller in engine fault-tolerant system theoretical research. Aiming at fault tolerant control problems, design the sliding mode controller. Isolation and estimates of fault signal are accomplished by fault-tolerant logic. In order to verify the effect of the fault tolerance by sliding mode controller, a passive fault tolerant of PID controller is designed to compare with the active fault tolerant control fault tolerance by sliding mode controller, results shows that the sliding mode fault tolerant controller could improve the flight performance in the sensor fault cases.