Abstract: In practical production of mines, the prediction of mine water inflow is of great significance for mine water prevention and control. Taking working face 1301 of Yuncheng coal mine as the research object, and without considering the influence of seasonal factors, ARIMA-the time series analysis model-is used to establish the functional relationship between mine water inflow and time, which proves that the time series of water inflow in working face 1301 of Yuncheng coal mine is affected by seasonal factors. Then, based on the principle of addition and decomposition of time series, the long-term trend, seasonal indexes, circulating factors and random parameters in the water inflow time series are separated and extracted, and the nonlinear regression correction model of water inflow prediction is established through applying the entropy method to determine the weight of each parameter. After that, the simulation results are compared with the water inflow by using ARIMA model ignoring the seasonal effect. The results show that the prediction of mine water inflow based on the non-linear time series of entropy weight is close to the measured water inflow, which verifies the accuracy of the method.