The occurrence of natural fractures is an important parameter affecting the normal stress and tangential stress of the fracture wall,and the stress state of natural fractures is an important factor determining the magnitude of the ground construction pressure.Accurate calculation and prediction of the construction pressure plays an important role in hydraulic fracturing design and construction.According to the natural fracture occurrence and formation stress state,based on the criterion for fracture initiation and extension,a calculation model of the ground construction pressure is established.Using the actual fractured reservoir for calculation and analysis,it is found that:in normal fault and strike-slip fault,the smaller the dip angle is,the smaller the construction pressure is,and the smaller the construction pressure is when the azimuth angle is closer to 0°,180°,360° in normal fault,and the smaller the construction pressure is when the azimuth angle is closer to 90°,270° in strike-slip fault.In the reverse fault,when the azimuth angle is 0°～45°,135°～225°,315°～360°,the larger the inclination angle,the larger the construction pressure,when the azimuth angle is 45°～135°,225°～315°,the larger the inclination angle,the smaller the construction pressure,under the same inclination angle,when the azimuth angle is in the range of 0°～90°,180°～270°,the construction pressure is reduced with the increase of azimuth angle.For the same inclination angle,when the azimuth angle is within the range of 90°～180° and 270°～360°,the construction pressure increases with the increase of azimuth angle.Accurate calculation of the construction pressure according to the occurrence of natural fractures can provide a theoretical basis for the selection and optimization of field fracturing parameters.
Coal calorific value calculation is crucial for power planning and mine exploration. Traditional lab techniques and empirical calculations often fail with high variability or limited data. We developed a machine learning framework for small-sample scenarios, combining compositional characteristics, data augmentation, and Bayesian hyperparameter adjustment to improve prediction accuracy. Four regression models (ANN, SVR, decision tree, and LightGBM) were trained on proximate, ultimate, and petrographic features to predict coal calorific value. Among these, the LightGBM model achieved the highest predictive performance with a test (R2 approximately 0.93) and the lowest error (RMSE approximately 0.19), outperforming the ANN (R2 approximately 0.91) and other models. Fixed carbon (FC) and volatile matter (VM) were the key predictors of calorific value, aligning with domain knowledge and model interpretability. The improved data-driven approach reliably estimates coal energy content from small samples, enabling evidence-based geological modelling and better drilling decisions for increased exploration efficiency. [Received: January 16, 2026; Accepted: May 8, 2026]
