Abstract: The development height of water conducting fracture zones (WCFZ) is an important basis for mine water prevention and safety mining. Due to the special sedimentary environment and process in the Shaanxi-Mongolia mining area,the height of WCFZ differs greatly from traditional understanding in this region. Considering the current lack of calculation methods for the height of WCFZ in the Shaanxi-Mongolia mining area,a simulation using UDEC was conducted on the development laws and evolutionary characteristics of WCFZ under different mining progress at the 2217 working face of the Yingpanhao coal mine in Ordos city. The relationship between large energy events and the development of WCFZ was analyzed,and a precise prediction method for regional WCFZ height was established based on the GA-SVR algorithm. The results showed that the characteristics of the thick sandstone layer of Zhuiluo formation give it the characteristics of a hard rock layer during overlying rock movement, and the fracture zone underneath it develops fully and stabilizes into a "saddle shape". Microseismic monitoring data can accurately reflect the height of WCFZ in the region,with a relative error of only 3. 90% in prediction. The coefficient of determination R2 of the established WCFZ height prediction model is 0. 81,the relative average error r is 14. 55%,and the prediction accuracy is significantly better than that of empirical formulas (R2 <0,r = 59. 57%). The research results have important practical significance for water conservation mining,safety production design,and mine water control in the Shaanxi-Mongolia mining area.

Key words: water conducting fracture zones,Shanxi-Mongolia mining area,deep mining,support vector machine,genetic algorithm,prediction method