| Anlin Coal Mine is a State-owned local coal mine, which has always used blasting mining with sublevel caving in instability coal seam No. 21 in recent years. Because coal seam was intruded by magmatic rock, the thickness of coal seam and magmatic rock both varied largely in different area of the minefield. Consequently, strata pressure behavior and top coal caving-ability were affected by them, and safely efficient production was also affected. With laboratory tests, field observations and numerical computation, thesis studied strata pressure behaviors, their parameters at coal face, influencing factors on movement abutment pressure and the predictive model for top coal caving-ability. Firstly, physical-mechanical parameters of surrounding rock in coal seam No. 21, were tested. Based on these and used numerical calculation and orthogonal design, thesis studied the relationship between above four factors and movement abutment pressure. The four factors were that thicknesses of coal seam and magmatic rock over coal seam, buried depth of coal seam and coefficient of horizontal pressure. Then thesis concluded primary and secondary orders of the four factors in light of affecting movement abutment pressure and the geological conditions in the context of the maximum and minimum movement abutment pressure. Secondly, based on field measurement of strata pressure, thesis obtained conclusions that movement rule of top coal, strata pressure behaviors and the parameters during the normal development of coal face. Strata pressure behaviors were obvious, roof weighting being differently along the dip direction of coal face, the mining with sublevel caving by individual hydraulic props was fitted for geological conditions of coal seam No.21, top coals fell hierarchically, their caving angle was 54.8o, and roof caving angle was 53.3o. Lastly, based on the induction of predictive index for top coal caving-ability and training samples, it is artificial neural network predictive models for top coal caving-ability, which was set up and trained. Then combined with orthogonal design, thesis made out that the law predicted by artificial neural network model was consistent with actual one. Using artificial neural network model, caving-ability of top coal at the coal face in Anlin Coal Mine belonged to normal type. These conclusions had crucial significances on advancing recovery ratio and safe production capacity at coal face in Anlin Coal Mine. |