Abstract: When there is noise in the time series signal of the natural circulation flow, wrong conclusion could be arrived in the calculation and analysis. In order to avoid this result, based on the experimental data signal of the supercritical water natural circulation flow, the denoising of the experimental flow signal is analyzed by selecting various wavelet basis functions. The calculation and comparison of the indicators showed that, the denoised signal value for the natural circulation flow after Dmey wavelet base function transformation are with miminum standard deviation and root-mean-square error (RMSE), maximum correlation coefficient and high noise-signal ratio. The wavelet basis functions is suitable for the analysis and treatment of the signal denoising of the experimental data of supercritical water natural circulation flow.