Research On Fault Diagnosis Of Diesel Engine High Pressure Common Rail System

In recent years,the time domain analysis methods represented by wavelet analysis,EMD（Empirical Mode Decomposition）,EEMD（Ensemble Empirical Mode Decomposition）,etc.are widely used in the analysis and diagnosis of diesel engine faults.In the field of eigenvalue identification,energy entropy,permutation entropy,singular values and time-domain features are applied to the extraction of eigenparameters.In the problem of classification combination,classification algorithms such as support vector machines and neural networks are applied in fault diagnosis.The methods of extracting the feature vectors of constructing vibration and vibroacoustic signals through time domain and frequency domain analysis and using classification algorithms for fault diagnosis are becoming more and more mature.However,the research on the fault diagnosis of high pressure common rail system under complex conditions is still not intensive,and some related problems extended by the fault diagnosis of high pressure common rail system have not been solved yet.In order to solve the fault problem of common rail system and related problems,this paper firstly obtains the rail pressure signal,flow signal,injection pressure signal,control signal and other related operation signals of common rail system under 3 operating conditions and 5operating states in 15 operating cases through bench test.In order to further improve the quality of the obtained signals,especially the rail pressure signals,this paper first uses a first-order lowpass analog filter to reduce the noise of the original rail pressure signals,and then uses fractal interpolation to fit the noise reduced rail pressure signals to obtain noise-free,high-quality rail pressure signals and verify the correctness and accuracy of this method.A one-dimensional simulation model was developed and validated using AMESim based on the data obtained from the bench test.For the fault diagnosis of high pressure common rail system,a variety of fault diagnosis strategies including two frequency domain analysis methods,two signal separation methods,four eigenvalue selection methods and two fault classification methods are studied in this paper,which are combined into a total of 86 fault diagnosis methods for high pressure common rail system.The diagnostic process of one of them is shown,and the advantages and disadvantages of different fault diagnosis strategies are analyzed to give reference to the selection of fault diagnosis for common rail system.On the other hand,almost most of the fault diagnosis methods need enough training data to build fault diagnosis models,and when the number of available training samples is small,fault diagnosis cannot be performed.In order to solve the problem of small sample fault diagnosis of high pressure common rail system,this paper investigates how to expand the training data by fitting samples when the training data is small,so as to solve the problem of small sample fault diagnosis of high pressure common rail system and further improve the possibility of applying the fault diagnosis method of high pressure common rail system in practical engineering.