Research On Fault Diagnosis Of Aeroengine Gas Path Based On Improved SVM

In recent years,the research on fault diagnosis of aero engine has become a hot topic.The abnormal operation of the engine is not desirable either from the perspective of engine operation or aircraft mission management.The condition of each component in the air-path system of the engine directly affects the overall performance and operation of the aircraft,so early detection of anomalies is crucial to the health management of the engine.In this paper,the fault database is expanded by means of mechanism modeling,aiming at the defect of less gas-path data.To solve the problem of random parameters of SVM model,GA is used to optimize LSSVM parameters.A mobile fault diagnosis platform is developed to meet the current needs.This paper deals with the mechanism modeling of aero-engine,fault data simulation,improved fault diagnosis algorithm of support vector machine and development of fault diagnosis platform.The main work of this paper is as follows:(1).The fault types and causes of aero-engine gas-path system are analyzed.Firstly,the paper analyzes the relationship between performance parameters and measurement parameters of gas circuit components.Secondly,a biaxial turbofan aero-engine was taken as the model object,and the mechanism model of aero system was established by using SIMULINK module in MATLAB.Finally,the established mechanism model is used to simulate the gas-path fault,so as to obtain the gas-path fault data.(2)Aiming at the characteristics of less and nonlinear data of aero-engine gas-path fault,LSSVM was used to diagnose the fault of aero-engine gas-path system,and genetic algorithm was used to optimize parameters of SVM and LSSVM respectively.Firstly,8 measurable parameters such as the speed of the low-pressure rotor,the speed of the high-pressure rotor,the outlet pressure of the fan,the outlet pressure of the compressor,the outlet pressure of the low-pressure turbine,the outlet pressure of the high-pressure turbine,the outlet temperature of the low-pressure turbine and the outlet temperature of the high-pressure turbine are selected as fault characteristics.Secondly,the simulation analysis of the improved algorithm is realized by using Python.Finally,by comparing GA-LSSVM with GA-SVM,LSSVM and SVM,the results show that GA-LSSVM is superior to other three algorithms in three aspects: diagnostic accuracy,noise strength and training time,and good simulation results are obtained.(3)This paper designs and implements a fault diagnosis platform based on WeChat applet aero-engine.Firstly,the feasibility analysis,functional requirement analysis and system framework of the mobile diagnosis platform are designed.Secondly,account management module,electronic chemical single management module,engine fault diagnosis module and engine fleet management module are designed and realized.Finally,the functions of the mobile diagnosis platform are tested.