Intelligent fault diagnosis of train bearings acoustic signals with Doppler shift based on the EMD and BPN

An approach to intelligent fault diagnosis of train bearings acoustic signals with Doppler shift is proposed in this paper, which based on the EMD and BPN without eliminating the Doppler shift effect, by extracting the fault characteristic information from train bearing acoustic signals. First, decompose the acoustic signals with Doppler shift into IMFs by EMD. Then, calculate the 8 VFs of IMFs and 10 TFs of original signals, take the 18 features as the input of BPN. After training the BPN, the result of test samples shows the proposed approach can discriminate different fault conditions of train bearing reliably and accurately.