To investigate the mechanism of hydraulic fracture propagation in coal seams with discontinuous natural fractures, an innovative finite element meshing scheme for modeling hydraulic fracturing was proposed. Hydraulic fracture propagation and interaction with discontinuous natural fracture networks in coal seams were modeled based on the cohesive element method. The hydraulic fracture network characteristics, the growth process of the secondary hydraulic fractures, the pore pressure distribution and the variation of bottomhole pressure were analyzed. The improved cohesive element method, which considers the leak-off and seepage behaviors of fracturing liquid, is capable of modeling hydraulic fracturing in naturally fractured formations. The results indicate that under high stress difference conditions, the hydraulic fracture network is spindle-shaped, and shows a multi-level branch structure. The ratio of secondary fracture total length to main fracture total length was 2.11~3.62, suggesting that the secondary fractures are an important part of the hydraulic fracture network in coal seams. In deep coal seams, the break pressure of discontinuous natural fractures mainly depends on the in-situ stress field and the direction of natural fractures. The mechanism of hydraulic fracture propagation in deep coal seams is significantly different from that in hard and tight rock layers.

In top-coal caving, the window control of hydraulic support is a key issue to the perfect economic benefit. The window is driven by the electro-hydraulic control system whose command is produced by the control model and the corresponding algorithm. However, the model of the window’s control is hard to establish, and the optimal policy of window action is impossible to calculate. This paper studies the issue theoretically and, based on the 3D simulation platform, proposes a deep reinforcement learning method to regulate the window action for top-coal caving. Then, the window control of top-coal caving is considered as the Markov decision process, for which the deep Q-network method of reinforcement learning is employed to regulate the window’s action effectively. In the deep Q-network, the reward of each step is set as the control criterion of the window action, and a four-layer fully connected neural network is used to approximate the optimal Q-value to get the optimal action of the window. The 3D simulation experiments validated the effectiveness of the proposed method that the reward of top-coal caving could increase to get a better economic benefit.