Application of Vibration Signal Analysis Method Based on Wavelet and LMD

Abstract

This paper proposes a vibration fault diagnosis method based on wavelet preprocessing and Local Mean Decomposition (LMD). In order to obtain time-frequency energy information accurately, flrstly, the original signal is transformed by wavelet. From the viewpoint of energy, using adaptive threshold for denoising, while most of the signal energy is reserved, the in∞uence of most noise is eliminated at the same time. Then, LMD is used to get the signal component which has deflnite physical sense and contains fault information. To identify the abnormal frequency components, the signal component is analyzed in time-frequency domain. Finally, the running state of hydro turbine can be diagnosed by analyzing the time-frequency information of each component signals. This method is applied in the No.1 hydro turbine of a power plant, the analysis shows that this method is efiective, and the running state of the turbine can be correct evaluated.