Fault Diagnosis For Aero-engine Based On Data-driven

Aero-engine,as the heart of aircraft,is generally working in the condition of high temperature,high speed,strong vibration,big stress,and it's working status varies over time.And high safety and reliability is also need.Due to the aero-engine manufacturing,technological level,and the later limitation of use,maintenance and management level,there are higher requirements on the economy of the engine.Aero-engine fault occupies considerable proportion in flight fault and often leads to catastrophic accidents in flight.Therefore,in order to make safe and efficient operation of the aero-engine,save the maintenance cost,it is necessary to understand the running status of the aero-engine,to grasp the law of its change,and the real-time state monitoring and fault diagnosis for the aero-engine.In this thesis,we present an approach for data-driven design of fault diagnosis system using the 40K aero-engine model.The core of this approach is only using the I/O data of aero-engine,whose exact mathematical model is unknown,to compute a series of related parameters,and then to construct a residual generator and a bank of isolation observers for the system fault diagnosis and isolation.Firstly,we present the data-driven design of fault detection and isolation systems for the linear time invariant system.And the proposed method is illustrated by a simulation study on 40K aero-engine model with actuator fault and sensor fault respectively.Then,considering the linear time invariant system with a time-varying parameter,propose a fault diagnosis scheme consists of an adaptive residual generator and a bank of isolation observers.The simulation of 40K aero-engine model is given in the end,which illustrates the effectiveness of the proposed scheme.Finally,the results of the dissertation are summarized and further research topics are pointed out.