Abstract: At present, there is no calculation model for predicting the pumping capacity of surface wells in mining stability areas in China, which is difficult to provide reliable basis for the pumping design of surface wells. Based on this, obtaining the single-day gas extraction capacity prediction model of surface wells by deeply digging the whole process extraction data of 5310 working face in Chengzhuang Mine, using the whole-process fitting comparison method to analyze the actual extraction data of surface wells under the influence of removing excess gas sources, comparing and verifing the productivity prediction model with the whole-process extraction data of surface wells in different working faces of Yuecheng Mine. The research results show that the daily pumping volume of surface wells in the mining stability area changes in an exponential function, and the correlation between the fitting data of the prediction model and the actual pumping data of different working faces at different stages of well YCCD-02 of Yuecheng Mine can reach more than 85%. At the same time, the constructed productivity prediction model predicted the coalbed methane extraction volume and the actual extraction volume of Yuecheng Mine in the later stage of well YCCD-02 reached 85.5%, which verified the accuracy of the prediction model, and can be used to guide the design of coalbed methane surface well pumping in the mining stability area and the scientific development of gas in the stable area.