Abstract: Aiming at the problem that the Fourier decomposition method (FDM) was difficult to extract fault features accurately in a strong noisy environment, taking the advantage of the FDM with orthogonality, completeness and locality,and the Teager energy operator (TEO) could highlight the impulse features in the signal,the following method was proposed.First, the FDM was used to search for the Fourier intrinsic mode component signals in a high-frequency to low-frequency search mode.Then,the valid mode signals selected by the kurtosis criterion and the correlation coefficient were used to reconstruct. Next,the TEO was used in the reconstructed signal to enhance the impulse features. Finally, the fault features were extracted in the spectrogram. This method was used to analyze analog and rolling bearing vibration signals, and bearing data provided by the University of Cincinnati was used for follow-up verification. The research results show that the method is significantly better than the FDM, which can accurately highlight the fault feature frequencies at 78Hz and 235Hz.The fault diagnosis of rolling bearing is realized by this method.

Key words: Fourier decomposition method (FDM);Teager energy operator (TEO);kurtosis criterion; correlation coefficient; rolling bearing