Research On Intelligent Diagnosis Method Of Diesel Engine Valve Clearance Abnormal And Misfire Fault

Diesel engines have the advantages of high thermal efficiency and large torque,and are widely used as power sources for tanks,ships and other equipment.However,due to long-term impact and wear,the valve clearance of diesel engines will be abnormally enlarged.Due to the accumulation of impurities,the fuel pipe will be blocked,and the diesel will have misfire failures.The occurrence of these failures will not only lead to the decline of diesel engine performance,but sometimes even cause serious disasters.Therefore,it is very important to diagnose the fault of diesel engine.There are many diesel engine parts,the working environment is harsh,and the working conditions are complex and changeable.Therefore,the diesel engine cylinder head vibration signal interference is strong,and how to dig out the signal characteristics under strong noise and multi-working condition interference is also a thorny issue.In addition,diesel engine failure data.Less,it takes a lot of manpower and material resources to obtain data through fault simulation experiments.In response to the above three issues,this article has done the following research work:(1)The signal decomposition method based on empirical wavelet transform avoids the modal aliasing problem of the traditional empirical modal decomposition method.The components can be filtered through correlation coefficients and time-domain features,and signal noise can be removed by noise reduction through a fixed threshold.(2)The cylinder head vibration signal is a non-stationary signal,which is suitable for time-frequency transformation through complex morlet wavelet base.Traditional feature extraction methods rely heavily on expert experience,so it is proposed to automatically extract vibration time-frequency image features through the Alexnet convolutional neural network.(3)Aiming at the problem of few diesel engine failure samples and complex working conditions,a fault diagnosis method based on migration learning is proposed.An adaptive layer is added to the network to quantitatively measure the difference between the existing working condition data and the new working condition of the diesel engine.The difference.(4)In terms of practical engineering applications,the effectiveness of the method proposed in this paper is verified by performing fault simulation experiments on actual engineering units.