The mining of layered soft bauxite under coal seams (BCS) will cause serious underground goaf disasters and surface Bayer process red mud (BRM) pollution. In order to realize the safe and efficient mining of BCS, the feasibility of recycling BRM as a backfilling aggregate was explored. A series of tests were conducted to prevent the pollution diversion of BRM from surface storage to underground goafs, and a numerical simulation analysis of the backfilling mining process was carried out based on FLAC^3D to protect the overlying coal seam. The results show that: under the action of encapsulation, solidification and inhibiting precipitation from cementitious materials, the leaching concentration of harmful substances in red mud-based cemented backfill (RCB) can be reduced 70% more than fresh BRM. Mining disturbance redistributes the in-situ stress field of overlying strata; normal backfilling can not only reduce the pressure stress of pillars, but also release the tensile stress in the roof and floor from +0.4956 MPa to −0.1992 MPa, effectively preventing roof subsidence. Since the creep damage process of past backfill will absorb and dissipate lots of energy, the disturbance range caused by backfill mining is controlled within 3 m, which is only 10% of the open-stope method.

Borehole breakouts significantly influence drilling operations’ efficiency and economics. Accurate evaluation of breakout size (angle and depth) can enhance drilling strategies and hold potential for in situ stress magnitude inversion. In this study, borehole breakout size is approached as a complex nonlinear problem with multiple inputs and outputs. Three hybrid multi-output models, integrating commonly used machine learning algorithms (artificial neural networks ANN, random forests RF, and Boost) with the Walrus optimization algorithm (WAOA) optimization techniques, are developed. Input features are determined through literature research (friction angle, cohesion, rock modulus, Poisson’s ratio, mud pressure, borehole radius, in situ stress), and 501 related datasets are collected to construct the borehole breakout size dataset. Model performance is assessed using the Pearson Correlation Coefficient (R²), Mean Absolute Error (MAE), Variance Accounted For (VAF), and Root Mean Squared Error (RMSE). Results indicate that WAOA-ANN exhibits excellent and stable prediction performance, particularly on the test set, outperforming the single-output ANN model. Additionally, SHAP sensitivity analysis conducted on the WAOA-ANN model reveals that maximum horizontal principal stress (${\sigma }_H$) is the most influential parameter in predicting both the angle and depth of borehole breakout. Combining the results of the studies and analyses conducted, WAOA-ANN is considered to be an effective hybrid multi-output model in the prediction of borehole breakout size.

Given the critical role of true triaxial strength assessment in underground rock and soil engineering design and construction, this study explores sandstone true triaxial strength using data-driven machine learning approaches. Fourteen distinct sandstone true triaxial test datasets were collected from the existing literature and randomly divided into training (70%) and testing (30%) sets. A Multilayer Perceptron (MLP) model was developed with uniaxial compressive strength (UCS, ${\sigma }_c$), intermediate principal stress (${\sigma }_2$), and minimum principal stress (${\sigma }_3$) as inputs and maximum principal stress (${\sigma }_1$) at failure as the output. The model was optimized using the Harris hawks optimization (HHO) algorithm to fine-tune hyperparameters. By adjusting the model structure and activation function characteristics, the final model was made continuously differentiable, enhancing its potential for numerical analysis applications. Four HHO-MLP models with different activation functions were trained and validated on the training set. Based on the comparison of prediction accuracy and meridian plane analysis, an HHO-MLP model with high predictive accuracy and meridional behavior consistent with theoretical trends was selected. Compared to five traditional strength criteria (Drucker–Prager, Hoek–Brown, Mogi–Coulomb, modified Lade, and modified Weibols–Cook), the optimized HHO-MLP model demonstrated superior predictive performance on both training and testing datasets. It successfully captured the complete strength variation in principal stress space, showing smooth and continuous failure envelopes on the meridian and deviatoric planes. These results underscore the model’s ability to generalize across different stress conditions, highlighting its potential as a powerful tool for predicting the true triaxial strength of sandstone in geotechnical engineering applications.