关键词: 充填开采, 底板破坏深度, 随机森林, 影响因素

Abstract: With the depletion of shallow coal resources in the eastern part of North China，resource extraction is carried out to greater depths，the geological conditions faced by coal seam mining have be⁃ come increasingly complex. Filling mining technology has gradually been valued and applied to the mining of close-range coal seams or coal seams with complex geological conditions. To further study the influence of coal seam filling mining on the failure depth of the floor，this paper collected the measured data of the failure depth of the floor from 43 unfilled mining faces and 11 filled mining faces. Firstly，the Pearson cor⁃ relation between the influencing factors and the failure depth of the floor was analyzed. Subsequently，the random forest model was used to predict the failure depth of the coal seam in unfilled mining and filled mining，and to study the influencing factors，and the prediction results were compared with those of the empirical formula. The results of the research show that，compared with empirical formulas，the random forest model has higher prediction accuracy，and filling mining significantly reduces the depth of coal seam floor damage. The characteristic importance analysis of random forest shows that for the bottom plate failure depth of the unfilled mining face，the weights of the influencing factors are respectively the inclined length of the working face (0.52) > the coal seam mining depth (0.26) > the coal seam inclination angle (0.22). For the filling mining face，the weights are respectively the coal seam dip angle (0.54) > the coal seam mining depth (0.38) > the inclined length of the coal mining face (0.08).

Key words: filling mining, depth of floor failure, random forest, influencing factors