Improved Empirical Mode Decomposition and Its Application to Fault Diagnosis of Train's Rolling Bearing

Empirical mode decomposition is used to obtain intrinsic mode functions of signals, while the results are not fully up to expectations, especially for multi-frequency signals. This paper proposes an improved empirical mode decomposition based on ideal analytical filter, which could increase amplitudes of components with high frequency. Analytical signal is obtained by Hilbert transform, and its Hilbert spectrum is got by the improved empirical mode decomposition. Intrinsic mode functions are completely obtained by this improved method in the emulation. The improved method is also applied to acoustic signal of rolling bearing, combined with Wavelet transform, and the diagnostic results are great. Results of both emulational tests and practical tests show the effectiveness of the improved empirical mode decomposition.