Due to the increase in the length of the mining face, the pressure characteristics and spatial distribution in fully-mechanized mining faces are different from those in typical mining faces, which leads to great challenges in roof management and the intelligent control of ultra-long mining faces. Taking the ultra-long mining face of a medium–thick coal seam in the northern Shaanxi mining area as an example and using field monitoring data for the working resistance of the hydraulic supports, a non-linear prediction method was used to extract the features of the dynamic data sequence of the working resistance of the hydraulic supports, and a deep learning method was used to establish a pressure prediction model for ultra-long mining faces based on the adaptive graph convolutional recurrent network (AGCRN) algorithm. In the proposed model, the supports in the fully mechanized mining face were regarded as the logic nodes of a topological structure, while the time-series resistance data for the supports were regarded as data nodes on a graph. The AGCRN model was used to determine the spatiotemporal relationship between the working resistance data of adjacent hydraulic supports, thereby improving the accuracy of the proposed model. The MAE and MAPE were employed as performance evaluation indices. When the node-embedding dimension was set to 10 and the time window was set to 16, the corresponding MAE and MAPE values of the prediction model were the minimum values. Compared with the reference models (i.e., the BP, GRU, and DCRNN models), the MAE and MAPE of the AGCRN model were 38.75% and 23.49% lower, respectively, indicating that the AGCRN model effectively demonstrates high accuracy in predicting the working resistance of supports. The AGCRN model was applied in the prediction of the working resistance of the supports of the ultra-long fully mechanized mining face. The results revealed that the working resistance of the supports in the lower and upper areas was relatively small along the strike, whereas the working resistance of the supports in the middle area was large, exhibiting a zoning pattern of “low-high-low” in terms of the average working resistance. In conclusion, the proposed model provides data references for the state of the hydraulic supports, pressure identification, and intelligent control of the ultra-long mining faces of the medium–thick coal seams in northern Shaanxi.

The main Jurassic coal seams of the Ordos Basin of northwest mining area have special hosting conditions and complex hydrogeological conditions, and the high-intensity coal mining of the coal seams is likely to cause groundwater loss and negative effects on the surface ecological environment. The research was aimed at predicting the height of the water-flowing fractured zone (WFFZ) in high-intensity coal mining in that area and gave instructions for avoiding water inrush accidents and realizing damage reduction mining during the actual mining procedure of the coal mine. In this study, 18 samples of the measured height of WFFZ in Jurassic coal seams were systematically collected. In the mining method, the ratio of the thickness of the hard rock to the thickness of the soft rock in the bedrock, buried depth, mining height, and working face length was selected as the input vectors, applied the sparrow search algorithm (SSA) to iteratively optimize the weights and thresholds of the Elman neural network (ENN), constructed an SSA-Elman neural network model. The results demonstrate that the improved SSA-Elman neural network model has higher accuracy in predicting the height of the WFFZ compared with traditional prediction algorithms. The results of this study help guide damage-reducing, water-preserving mining of the middle-deep buried Jurassic coal seams in the northwest mining areas.

The back-filling body in the gob-side entry retaining is subject to continuous disturbance due to repeated mining. In this study, uniaxial and cyclical loading tests of back-filling concrete samples were carried out under laboratory conditions to study damage evolution characteristics with respect to microscopic hydration, deformation properties, and energy evolution. The results showed that, due to the difference in the gradation of coarse and fine aggregates, the cemented structure was relatively loose, and the primary failure modes under cyclical loading were tensile and shearing failure, which significantly decreased its strength. With an increasing number of loadings, a hysteresis loop appeared for the axial strain, and the area showed a pattern of decrease–stabilization–increase. This trend, to a certain extent, reflected the evolution of the cracks in the back-filling concrete samples. The axial, radial, and volumetric plastic strain curves of the back-filling concrete samples showed a “U” shape. The plastic strain changed in three stages, i.e., a rapid decrease, stabilization, and a rapid increase. A damage parameter was defined according to the plastic strain increment to accurately characterize the staged failure of the samples. The plastic strain and energy dissipation of the samples were precursors to sample failure. Prior to the failure of the back-filling samples, the amount and speed of change of both the plastic strain and energy parameters increased significantly. Understanding the characteristics of plastic strain, damage evolution, and energy dissipation rate of the back-filling samples are of great reference value for realizing real-time monitoring of back-filling concrete in the gob-side entry retaining and providing early warning of failure.