In view of the problem of the unstable vibration and noise interference in the rolling bearing fault signal acquisition in a reducer of the hoist of miter gate,a fault diagnosis method based on wavelet packet decomposition and power spectrum analysis were put forward based on wavelet threshold denoising for the diagnosis of rolling bearings.In this method,the fault signal was decomposed by wavelet function,and the decomposition coefficient was processed with threshold.The denoised signal was obtained by inverse wavelet transform of the decomposed coefficients.Then,the denoised signal was decomposed by wavelet packet,and the node with energy concentration was found,and the Hilbert envelope was demodulated and the power spectrum of its Hilbert envelope was obtained,so as to extract the fault feature information.The application example shows that the simulation signal and the fault diagnosis method for the reducer in the hoist of miter gate mechanical hoist can reduce the signal noise and interference,and extract the fault feature information.