Experimental study on cascade optimization of geological models in intelligent mining transparency working face

Coal mine intelligent mining technology and equipment are not adaptable to geological conditions, and it is urgent to build a high-precision and transparent coal seam geological model under various complex geological conditions.Taking a mine with complicated geological conditions in Shanxi as an example, the more conditions in Working Faces XY-S such as collapse columns, faults, folds, etc.were selected to discuss cascade modelling and optimizing, and the black box model at the design stage was constructed progressively by using the geological data obtained at different exploration and production stages.The accuracy of above four models has been verified according to the measured data which consisted of 7 400 m excavated roadways and 1 470 m mined area.The compared indicators includes the prediction accuracy of coal seam floor and thickness, and the explored geological structures.The test results show that: ① modeling error of coal seam floor: black box model 10～20 m(only drilling data is available), 5～10 m(drilling + 3D seismic), the error ranges of gray box model and white box model are 0～5 m, and the range for transparent model is 0～1.0 m; ②The degree of control of faults and subsidence columns: three faults with a gap difference more than 1.5 m explained by in-seam seismic(ISS)are verified reliably.The interpretation accuracy of the subsidence column with a diameter no more than 20 m is 75% on average, but the ranges of subsidence column detected by ISS tend to be larger than the actual ranges.③coal thickness prediction error:the average thickness of the main coal seam is 2.70 m.The maximum error of the coal thickness prediction of the black box, gray box and white box model is 1.5 m and the mean square error is about 0.5 m.The transparent model’s coal thickness prediction error is less than 0.30 m, but there are few empirical statistical points.According to the idea of cascade geological modeling,the accuracy of white box model can basically satisty the demands of adaptive cutting of intelligent coal mining.In order to improve the accuracy of longwall panel model, it is urgent to develop intelligent detection with, geological radar in hole, and video coal and rock identification with new technologies and new equipment, which can realize high-precision three-dimensional geological modeling of the working face and provide reliable geological guarantee for coal mine intelligent mining.