Abstract
As a main source of power, diesel engines are widely used in large mechanical systems. Fire failure is a kind of common fault condition, which seriously affects the power and economy of the diesel engine. Previously, scholars mostly used single-channel signal to diagnose the misfire fault of the diesel engine. However, the single-channel signal has limitations in reflecting the information of fault. A novel fault diagnosis method based on MEMD and dispersion entropy is proposed in this paper. Firstly, the multichannel vibration signal of the diesel engine cylinder head is decomposed by multivariate empirical mode decomposition (MEMD), which obtains the IMF component groups with the same frequency in the same order. Then, the IMF component with a large correlation coefficient with the original signal in each group is selected to reconstruct new signal, and dispersion entropy (DE) of the reconstructed signal is calculated as a fault feature vector. Finally, the fault feature vector is input into the support vector machine (SVM) for misfire fault classification. Compared with the other three methods, the results show that the diagnosis method proposed in this paper can effectively extract the fault features and accurately identify the fault type, which is superior to the comparison method.
Highlights
Diesel engine is the main power mechanism for a heavy equipment, whose working conditions determine the reliability and safety of the whole system
When a cylinder of diesel engine fails, the complexity of vibration signal on cylinder cover changes due to the evenness of the cylinder burst. erefore, this paper proposes to extract feature by dispersion entropy from multichannel signal
Support vector machine (SVM) is a machine learning method that deals with pattern recognition, probability estimation, and other issues, which has advantages when dealing with small sample data classification. e kernel function is the main factor determining the performance of support vector machine (SVM); RBF kernel function can effectively deal with nonlinear problems, so this paper selects RBF kernel function for diagnosis recognition
Summary
Diesel engine is the main power mechanism for a heavy equipment, whose working conditions determine the reliability and safety of the whole system. Considering the nonlinear nonstationary characteristics of diesel engine vibration signal, traditional signal methods cannot effectively extract fault feature information. E multivariate signal can be divided into multiple one-dimensional signals after processed by EMD, which may cause the problem of different amount of IMF components obtained after decomposition, or inconsistent frequency corresponding to the same order IMF components. It does not make any sense for signal analysis. For the three-dimensional signal, the trivariate EMD method was proposed by Ur Rehman and Mandic [16], which projects the multivariate signal on the threedimensional spherical direction vector and calculates the mean value.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
