Exploration and prediction of high pressure dynamic water hidden collapse column in coal mines

Abstract

Hidden collapse column associated with high pressure dynamic water is a main cause of major water inrush accidents in North China type coal fields. Taking the structural abnormality area discovered in 11603 working face of Daizhuang Coal Mine as an example, underground three-dimensional high-density electrical method, advanced exploration of underground drilling and curtain grouting were used to detect the existence of collapse column, and analyzed the water conductivity of collapse columns based on the hydraulic connection analysis of the 13th limestone and Ordovician limestone aquifers. Finally, it is determined that this abnormal area is a strong water filling collapse column originating from the upper Ordovician strata runoff zone (inferred to be within a range of 30 to 100 m below the Ordovician limestone top interface), developed to a height of 12th limestone. Based on the fact that the water yield and water pressure of underground directional drilling, the grouting pressure of curtain grouting, and the amount of cement injected are external quantitative factors that reflect the existence of hidden karst collapse columns during the process of detecting hidden karst collapse columns, and in combination with the feature that deep learning can fully independently learn abstract knowledge expression, a prediction model based on convolutional neural networks is constructed. According to the established network model, it was found that among the 12 sets of actual measurement data, only one data point indicated the absence of a collapse column. The prediction accuracy reached 91.6%, which meets the practical needs.