Abstract: In order to understand the intrinsic mechanism,non-linearity and chaotic behavior of gas-liquid two-phase bubbly flow in horizontal pipe,the pressure fluctuation signals with different flow rates were collected with a high speed collector.The pressure fluctuation signals with different flow rates denoised by using the wavelet transform modulus maxima method were firstly decomposed to 1-7 scale detail signals and scale approximation and then evaluated by using Hurst analysis method based on Daubechies second order wavelet.By analyzing the profiles of different scale signals of the pressure fluctuation signals,different fractal structures at different scales were found.The results show that 1-2 scale detail signals have only one Hurst exponent,and less than 0.5,reflecting the micro-scale interaction between bubbles and bubbles;3-7 scale detail signals have two Hurst exponents,one smaller than 0.5 and one larger than 0.5,reflecting the micro-scale interaction of the liquid phase and bubbles,respectively;8-9 scale detail signals and 9 scale approximation signal have only Hurst exponent which larger than 0.5,reflecting the macro-scale interaction of the gas-liquid two-phase flow system and the interaction between the wall.The energy profiles of different scale signals indicate that pressure fluctuation signals mainly reflect the micro-scale interaction between bubbles and bubbles.