Fault diagnosis of internal combustion engine using Empirical Mode Decomposition

In this paper, a signal decomposition and feature extraction technique for the fault diagnosis of internal combustion engine, based on empirical mode decomposition (EMD) is presented. Vibration signal measured from a faulty engine is decomposed into a number of intrinsic mode functions (IMFs), with each IMF corresponding to a specific range of frequency components embedded in the vibration signal. However, in our application we found that all the IMFs are not useful to reveal the vibration signal characteristics due to the effect of noise. Hence cumulative mode function (CMF) is presented. With CMF, the most representative IMFs are combined to obtain an oscillation mode representing signal features more accurately. Two criteria, the energy measure and correlation measure, are investigated to determine the most representative IMFs for extracting fault induced characteristics features out of vibration signal. Statistical parameters, shape factor, crest factor, etc., of the envelope spectrum of CMF are investigated as an indicator for the presence of the fault.