Fault Diagnosis Of Diesel Engine Based On EMD Decomposition

As a kind of complicated power machine, diesel engine is widely used among ships, locomotives, automobiles and generating sets, and its running state influences directly on security and reliability of the whole power system. Vibration is generated while the engine works, and the state information of the internal parts is transmitted to the surface vibration by some way, so it's an effective way to take diesel fault diagnosis by vibration signal. Fault diagnosis technology is taken as the research object, features extraction and fault diagnosis method is researched in-depth based on the cylinder head vibration signals in this paper.Following aspects are included in this paper:first, the features of the cylinder head vibration signals are analyzed by theory and experiment, the features of the vibration source are revealed, the production mechanism and transmission routes of the fault are analyzed, and the time domain-analysis,frequency-analysis of the diesel faults are discussed. Second, EMD method is used to decompose the cylinder head vibration signals, physical significance of the IMF is analyzed. HHT marginal spectrum method is taken to analyze the features of the faults. Third, use time sequence analysis method and fractal theory to analyze the features of the vibration, calculate the AR model parameters and correlation dimension of each IMF, from the calculated results, some relationship between feature vector and work condition is verified, so the features that can reveal the fault and that can not are separated. Fourth, the nature of working condition identification is state classification, AR model parameters and correlation dimension are trained by neurotic network and support vector machine to identify the diesel working states. At the same time, the feasibility of this method is verified; also, from comparation we can see that the rate of identification by support vector machine is higher than neurotic network; SVM is more useful to analyze small number of samples.By theory analysis and calculation, fault mechanism and fault diagnosis methods are researched in-depth under abnormal gas valve clearance and oil break condition, several conclusions and methods that have important project value are obtained, which have very important reference value on diesel fault diagnosis.