Fault Feature Extraction of Diesel Engine by Using Ensemble EMD and Morphological Fractal Dimension

Abstract

The vibration signals generated by the diesel engine can be regarded as a typical nonlinear signal. The fractal geometrical theory offers an effective tool to describe the nonlinearity of such a signal. Aiming at the characteristics of nonlinearity and low noise–signal ratio of engine vibration signals, in this paper, the condition monitoring of diesel engines based on ensemble empirical mode decomposition (EMD) and morphological fractal dimension is studied. First, vibration signal is decomposed into a set of intrinsic mode functions (IMFs) by ensemble EMD in order to suppress the noise interference and obtain the fault feature information of the characteristic IMF. Then the fractal dimension of the characteristic IMF is calculated and used to evaluate the fault type of the engine. The analysis of diesel engine vibration signals at different states has been done. It is noted that the ensemble EMD can effectively separate the characteristic components from engine vibration signals, and the fractal dimension can quantitatively describe the geometric characteristics of the engine vibration signals. The studies show that this method can effectively extract the fault feature of diesel engines.