Research On Engine Misfire Fault Diagnosis Based On Vibration Signal Analysis

As the power source of a car,the probability of engine's failure is very large,and maintenance costs are relatively high,maintenance time is longer.The single cylinder misfire fault is a kind of common faults in use of engine.It will be significative if the fault location and fault components can be determined under the condition of no disintegration.The method of extracting fault characteristics from the vibration signal is simple,so it is easy to realize the fault diagnosis,even for diagnosis without disintegration.Neural network is applied to fault diagnosis of electronic controlled engines and can identify the fault types,and the accuracy is high.Based on the analysis of the present research situation about fault diagnosis at home and abroad,a diagnosis method of electronic control engine misfired fault based on vibration signal is presented.Based on the basic structure and working principle of the engine,the causes of the engine failure are determined.The vibration of the engine is studied.The vibration excitation source and the vibration propagation path are analyzed.Analysis methods of engine vibration signals are studied,including time domain analysis and frequency domain analysis.Wavelet analysis method is mainly studied.Then the engine failure test program is designed,and the vibration signals in the single cylinder misfired and normal state are collected on the DN4G15 electronic control engine.Then,the vibration signals of the engine cylinder head are analyzed in time domain and frequency domain.The wavelet packet analysis is used to reconstruct the vibration signal,and the band energy of the decomposed node is extracted.Finally,according to the results of vibration signal analyzed,the characteristics of engine failure are extracted,and the genetic neural network model of electronic control engine fault diagnosis is established.The extracted samples are trained and tested by neural network.Some samples are randomly selected for diagnosis.The results show that the genetic neural network can make the correct judgment of most samples,and the diagnostic accuracy is higher than BP neural network,which shows that the method is feasible.The research in this paper can provide reference for engine fault diagnosis and has certain practical significance.