Abstract: In order to improve the prediction accuracy of coal mine dustproof water consumption, a combined model was presented based on wavelet analysis, gray prediction model (GM (1, 1)) and auto-regressive moving average (ARMA(p, q)) model.The time series of water consumption was decomposed into different scales by wavelet analysis, then the low frequency signal and the high frequency signal were predicted by GM (1, 1) and ARMA(p, q), finally the result was obtained by wavelet reconstruction. Taking Linnancang Coal Mine as the research background, the combined model was used to predict the water consumption in each month of 2014, the relative error of residual test was no more than 2.5% compared with the actual data.The results showed that the coal mine dustproof water consumption increases slowly year by year, with periodic changes every year, and the prediction model based on the combination of wavelet analysis and GM (1, 1) -ARMA(p, q) has high precision.