Abstract: Experimental research on the pressure difference fluctuation signal was conducted in the channel of a horizontal circular pipe with a diameter of 3.0 mm, using air and water as the working medium. According to the pressure difference fluctuation signal diagram and the flow pattern diagram taken by the high-speed camera, the chaotic dynamic analysis of the pressure difference fluctuation signal was carried out by using the phase space reconstruction and Lyapunov index discrimination methods. Then the Volterra adaptive short-term prediction of the pressure difference fluctuation signal was carried out. The experimental results show that the attractor diagram obtained by the chaos analysis can show the flow characteristics more accurately. The Volterra adaptive prediction model has the relative errors of 1.86%, 0.71%, 3.90% and 2.49% for the prediction of the pressure difference time series of the annular flow, the layered flow, the intermittent flow and the slug flow respectively, which can effectively make short-term prediction of pressure difference time series of the gas-liquid two-phase flow in the pipeline.