In order to evaluate the denoising effect of the mechanical structure monitoring signal， the concept of information entropy is introduced into the dual tree complex wavelet decomposition. Adaptive denoising method of the monitoring signal combined dual-tree complex wavelet transform （DT-CWT） and sample entropy （SE） is proposed. Firstly， the multi-level decomposition of the noisy signal is carried out by using the dual tree complex wavelet transform. Secondly， the sample entropy of each scale detail component and the sample entropy errand of detail component in adjacent scales are calculated. Comparing the sample entropy errand， the optimal decomposition level of dual tree complex wavelet is determined. Finally， the denoising threshold of wavelet coefficients of each level is determined according to the change of sample entropy in each scale， reconstructing denoised wavelet coefficients to realize adaptive denoising. The simulation analysis and experimental results show that the method is more thorough in denoising the monitoring signal， and more complete in preserving the original information. The denoising effect is better than the traditional wavelet threshold denoising method.