A Study On Automobile Engine Fault Diagnosis Based On Wavelet Neural Network

Engine is the core parts of the automobile, improving safety and reliability of engineoperation, and finding out failures and troubleshooting timely and effectively, can reducemaintenance costs of machinery, avoid the occurrence of major accidents. It has veryimportant significance on the research of intelligent fault diagnosis system of automobileengine. The engine is a typical reciprocating machinery, its complicated structure determinesthe condition monitoring and fault diagnosis of it is very difficult.In this context,this paper gives a fault diagnostic method based on engine vibrationsignal by using wavelet packet and neural network.Wavelet packet is capable ofdistinguishing each wave frequency to the desired level of detail, it is very suitable fornon-stationary signal processing. It can not only resolve low-frequency signals, but it canalso do the same to the high-frequency signals. Artificial neural network is a system of largescale distributing parallel processing, it has the features of self-organization, self-learning,self-adapting and non-linear dynamic processing, it has the broad prospect to solve complexnonlinear problems. So the method of combining wavelet packet and neural network in faultdiagnosis will be widely used.The fault diagnosis of engine is based on the analysis ofresearch status at home and abroad, the basic structure and working principle of the engineare analyzed, enumerates some main engine fault and the causes of fault, and analyze thefault characteristics and the vibration signal characteristics of the engine, which provides thetheoretical basis for the following research of engine fault diagnosis. The intake and exhaustsystem is an important part of the engine, its main function is the inhalation of air that engineneeds and the exhaust gas produced by the combustion of the engine, so the engine’s airintake system and exhaust system fault will have a very big impact on the car engine runningand working states.The fault diagnosis method of engine in this paper is:collect the vibration signal of theengine under three conditions,including normal working state,the intake pipe and exhaustpipe blockage. usethe wavelet packet threshold noise reduction methods to reduce thecollected signal noise; analyze the vibration signal of engine cylinder head in time domainand frequency domain; apply wavelet packet analysis to extract eigenvectors, put thesamples to BP neural network for training and the results achieves the desired effect. Verified by experiment on the proposed method, the results show that using the methodof combining wavelet packet analysis and the neural network is able to diagnose the fault ofautomobile engine correctly, this provides a new technical method for the engine conditionmonitoring and fault diagnosis,this method has the broad application prospect.