Locomotive fault diagnosis based on local mean decomposition demodulating approach

The identification of the locomotive bogies incipient faults is of great meaning to increase the heavy freight locomotive transport capacity and prevent severe accidents. According to the vibration signals of locomotive, a fault diagnosis method based on local mean decomposition (LMD) demodulating approach is proposed, which decomposes the signals adaptively into a set of product functions. Decomposition and demodulation are implemented together during the process. Compared with Hilbert Huang transform, LMD method calculates instantaneous frequency bypassing the Hilbert transform and involves no demodulation error of windowing effect. Breaking down the limitations of Bedrosian theorem and Nuttall theorem, the negative frequency does not exist. For the reason that the local mean and envelope are obtained by using sliding averaging, there are no phenomena about over enveloping, under enveloping and breakpoint effect. The method has been successfully applied in fault diagnosis to rolling bearing and gear of locomotive bogies. Compared with the results of EMD, it shows that LMD decomposes signals into demodulation components as much as possible and gets very suitable for processing multi-components vibration signals.