Machinery Fault Diagnosis Based On Hilbert-huang Transform

Rotating machinery is the key to national economic production equipment, so it is necessary for its operation status monitoring and fault diagnosis. The use of machinery and equipment vibration signals for fault diagnosis is a more effective and economical method. When the mechanical equipment fails, there are mostly non-stationary and nonlinear vibration signals, which the traditional signal analysis method is hard to deal with. Therefore, it is necessary for time-frequency analysis method research.In this paper, the necessity of using the time-frequency analysis methods such as the short time Fourier transform, the Wigner-Ville distribution and the wavelet transform, have been introduced. Meanwhile a new time-frequency analysis method--Hilbert-Huang Transform have been detail studied and compared with other typical time-frequency analysis methods. Because of the existence of the endpoint effects in Hilbert-Huang Transform, a continuation of the signal sequence in extension mirror extreme attenuation method has been presented and the effective suppression of the endpoint effect have been verified. At the same time, the impact of noise on the Hilbert-Huang Transform was studied, some improvements in the processing have been proposed.The failure characteristics of rolling bearings were studied. The experiment of rolling bearings includes inner crack, outer crack and health was made and the vibration signals acquisition was saved and analyzed by the EMD-base fault diagnosis method. As a result, the effectiveness of the envelope of EMD-based spectral analysis was verified. Meanwhile, the roller bearing fault was studied by the intrinsic mode of energy entropy.