Research On Vibration Source Localization And Fault Diagnosis Method Of Aero-engine Based On JWTS Method

Aeroengine fault diagnosis is very important to ensure the safe operation of aeroengine,this paper combines with blind separation technology,autoregressive model technology,wavelet decomposition technology to carry out aero-engine fault diagnosis,including the estimation of the main vibration source of the aero-engine and the extraction of the main components of vibration source,the extraction of the aero-engine fault pulse signal,the fault diagnosis based on the model,etc.The main contributions of this paper are as follows:Combined with the characteristics of the vibration spectrum of the rotary shaft,a joint wavelet transform and time synchronous averaged blind separation method(shortly for JWTS)is proposed.In the view of classical blind separation method cannot effectively separate the mixture signals measured on aero-engine casing,this paper combined with the characteristics of vibration spectrum of aero-engine vibration source,on the basis of predecessors’ research,proposed the JWTS method.This proposed method can effective extract the main vibration sources within the aero-engine,including high-pressure rotor source,low-pressure rotor source,combustion chamber source,which is tested by the real aero-engine mixture signal.The application of JWTS to aero-engine fault diagnosis can help to improve the accuracy of fault diagnosis.Proposed a joint hard threshold method and hybrid threshold method(shortly for hard-hybrid method)to extracting impulse signals.When the mixed signal contains only Gaussian noise and impulse signal,the hybrid threshold method can effectively extract the pulse components.However,when the mixed signal contains Gaussian noise,periodic signal and pulse signal,extracting the impulse signal through the hybrid threshold method has poor results due to noise and periodic signal interference.If the Gaussian noise component in the mixed signal is first passed out by the hard threshold method,the impulse signal can be clearly extracted by the mixed threshold method.The method is verified by both the analog signal and the test signal.This method is complementary to the blind extraction algorithm proposed in this paper.Because the blind separation algorithm proposed in this paper can only extract the sub-harmonic components of the harmonics corresponding to the fundamental frequency and cannot extract the impulse signal,the method can effectively supplement the JWTS algorithm.Proposed a fault diagnosis method based on model Env-AR-ARX model.According to the observation of the vibration signal in the casing,the autoregressive black box model of the aero-engine is established,and the residuals of the model output are classified by the variance analysis to diagnose the fault and improve the probability and accuracy of the fault diagnosis.The method is verified by the measured vibration signal.