To address insufficient dynamic characterization and unsystematic multi-index integration in tailings dam risk assessment, a comprehensive measurement method coupling numerical simulation with uncertain measure theory is proposed. A pyrite tailings reservoir was taken as the case study. FLOW-3D was used to perform three‑dimensional water‑sediment coupled dam‑break simulations, from which key dynamic parameters (maximum deposition depth and peak flow velocity) were measured. The 3DMine model was coupled to quantify inundation extent and economic losses. A four‑index evaluation system was constructed, and linear uncertain measure functions were applied for risk classification under three dam‑height scenarios (60 m, 63.5 m, and 65 m). Measured results show that, as dam height increases, breach initiation advances from 39 s to 28 s, maximum deposition depth rises from 4.35 m to 5.26 m, and peak velocity increases from 14.88 m/s to 15.22 m/s. The 370 m side drift remains safe, while the 352 m side drift, concentrator plant, and office area become inundated. Risk classification yields Grade III/II for 60 m, Grade IV/II for 63.5 m, and Grade IV for 65 m. The proposed measurement‑based methodology enables systematic quantitative evaluation and supports dam‑height optimization and downstream disaster mitigation.

Uniaxial compressive strength (UCS) is a fundamental parameter for rock engineering design and stability assessment, but direct laboratory testing is costly, time-consuming, and often difficult for weak or fractured rocks. To improve predictive accuracy while preserving mechanical interpretability, this study proposes a Gated Empirical-Law LightGBM model (GEL-LightGBM). The framework embeds three representative rock-strength priors, including point-load strength, multi-index strength, and porosity-degradation relationships, as empirical-law experts. A sample-adaptive gating mechanism dynamically assigns its contributions for different rock states, while a controlled residual corrector captures nonlinear deviations between empirical estimates and measured UCS. Using 344 published rock-mechanics samples, porosity, Schmidt rebound hardness, P-wave velocity, and point-load strength index were used as predictors. GEL-LightGBM outperformed LightGBM, XGBoost, random forest, MLP, CNN, SVR, and BPNN, achieving a testing R2 of 0.9790 and an RMSE of 7.5623 MPa. SHAP analysis identified porosity as the dominant factor, contributing 49.0%, followed by rebound hardness (32.2%) and P-wave velocity (17.2%). The strongest interaction occurred between porosity and rebound hardness (2.31 MPa). These findings indicate that GEL-LightGBM provides accurate, stable, and physically interpretable UCS prediction for heterogeneous rock datasets.

Deep roadway excavation in water-rich shale formations faces coupled challenges of long-term water saturation and cyclic blasting dynamic disturbance, yet the true triaxial mechanical behavior and coupled damage mechanisms of water-saturated shale under such conditions remain unclear. This study converts field blasting loads into laboratory stress paths via on-site monitoring, Fourier transform processing, and Miner's rule derivation, and conducts true triaxial fluid-structure coupling tests on shale specimens with varying saturation durations and pore water pressures, integrated with acoustic emission (AE) monitoring, post-test computed tomography (CT) scanning, 3DEC numerical simulation, analytical modeling, and neural network verification. The results show that pore water pressure acts as a damage amplifier, accelerating strain accumulation and damage evolution, while water saturation preconditions the microstructure: it transforms failure modes from localized brittle fracture to distributed ductile shear damage, and suppresses the permeability threshold of natural specimens via clay swelling-induced fracture network modification. Notably, a maximum damage point is identified at 24 h of water saturation, where the synergistic degradation of saturation-induced weakening and disturbance-induced damage peaks across all pore water pressure conditions. The developed neural network model achieves high accuracy in predicting post-disturbance mechanical properties. This work provides critical theoretical support for stability control and support design of water-rich shale roadways during blasting excavation.

The rheological behavior of paste in mine backfilling systems is governed by multiple coupled mechanisms, including particulate structure evolution, time-dependent effects, spatially heterogeneous flow, and scale dependence. As a result, its macroscopic response cannot be adequately described by a single material parameter or purely local constitutive relations. Although significant progress has been made in experimental characterization and empirical modeling, rheological parameters reported under different conditions remain difficult to reconcile, highlighting the limitations of existing models in capturing structural evolution and nonlocal effects. This review provides a concise synthesis of current advances in paste rheology for mine backfilling applications, with emphasis on yield behavior, shear-rate-dependent nonlinear flow response, thixotropy, and shear history effects. The applicability and limitations of commonly used rheological models, including the Bingham and Herschel-Bulkley models, are critically examined. Key factors influencing paste rheology-such as particle gradation, temperature, and chemical additives-are discussed from a structure-controlled perspective. Finally, physics-constrained data-driven approaches are highlighted as a promising direction for improving the description and prediction of complex rheological behavior. Overall, this review emphasizes the need to balance experimental observability, model simplicity, and physical consistency, and highlights the importance of linking microstructural mechanisms, scale effects, and macroscopic rheological response to establish more unified and engineering-relevant frameworks for paste rheology in mine backfilling systems.

Rock masses with certain shear strength are fundamental for ensuring the safety and stability of geotechnical engineering projects for geological disaster prevention. However, serrated jointed rock masses exhibit complex geometries and nonlinear mechanical properties, making accurate predictions of their shear strength challenging. To address this, an innovative machine learning-based prediction framework is proposed, integrating swarm intelligence optimization techniques with explainable data-driven methods to enhance prediction accuracy and reduce costs. This study utilizes experimental data of serrated jointed rock masses, covering key parameters such as internal friction angle, joint normal stress, ratio of normal stress to intact rock tensile strength, joint inclination, and shear strength. Based on this, various models were constructed, including Support Vector Regression (SVR), Backpropagation Neural Network (BPNN), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). Furthermore, the models were optimized using Sparrow Search Algorithm (SSA), Chameleon Optimization Algorithm (CSA), Snake Optimization Algorithm (SO), and Kepler Optimization Algorithm (KOA). Results of statistical performance indicators showed that the KOA-XGBoost model performed best in predicting both the training and testing sets (R2 of 0.992, RMSE of 0.197 and 0.239), significantly outperforming other comparative models (R2 of 0.895 to 0.988, RMSE of 0.248 to 0.808). TreeSHAP analysis revealed that joint normal stress and joint inclination (with a cumulative importance score exceeding 0.6) were the most critical factors influencing shear strength. The findings provide an effective solution for ensuring the safety and stability of geotechnical projects involving serrated jointed rock masses.

Supercritical carbon dioxide (SC-CO2) jetting has emerged as a promising technique for rock fracturing due to its superior physical properties such as low viscosity, high diffusivity, and zero surface tension. However, the complex interaction mechanisms between SC-CO2 jets and heterogeneous rock media remain inadequately understood. In this study, a coupled Smooth Particle Hydrodynamics–Finite Element Method (SPH-FEM) framework is established to simulate the dynamic fracturing process of rocks under SC-CO2 jet impact. The Riedel–Hiermaier–Thoma (RHT) constitutive model is incorporated to describe the nonlinear damage evolution of brittle rocks, and key material parameters are calibrated via sensitivity analysis and SHPB experimental validation. A series of numerical simulations are performed to investigate the effects of jet standoff distance, jet velocity, and rock lithology (marble, granite, red sandstone) on fracturing efficiency. Damage area, damage volume, and a novel metric—block size distribution—are employed to quantify the fracturing quality from both macro and meso scales. The results indicate that SC-CO2 jets outperform conventional water jets in creating more extensive and homogeneous fracture networks. An optimal standoff distance of 1–2 cm and a velocity threshold of 0.2 cm/μs are identified for maximum fracturing efficiency in marble. Furthermore, smaller block sizes are achieved under higher velocities, indicating a more complete and efficient rock fragmentation process. This study provides a comprehensive numerical insight into SC-CO2 jet-induced rock failure and offers theoretical guidance for optimizing green and water-free rock fracturing techniques in complex geological environments.

This study introduces an innovative application of a hybrid empirical-data-driven neural network (HEDDNN) for regression-based prediction of the air-entry value (AEV) in unsaturated soils, offering a novel approach to predicting soil hydraulic properties crucial for understanding water movement, rainfall infiltration, and groundwater recharge in hydrology and engineering geology. AEV, a critical parameter in unsaturated soil water migration, significantly influences processes such as soil erosion, surface runoff, landslide initiation, and groundwater recharge. A comprehensive database of 214 representative soil samples from diverse geological origins and soil types was constructed, ensuring broad model applicability. The HEDDNN model achieved remarkable predictive performance on the test dataset, with an R2 value of 0.980 and root mean square error (RMSE) of 2.294 kPa, outperforming traditional machine learning models like LightGBM [light gradient-boosting machine] and support vector regression (SVR). Feature importance analysis identified fines content, initial water content, and plasticity index as key AEV predictors. These findings underscore AEV's importance in modeling soil hydraulic behavior across diverse depositional environments and its impact on geological phenomena like slope stability and subsurface water flow. This dual-driven HEDDNN framework not only enhances AEV prediction accuracy but also bridges the gap between physics-based and data-driven modeling, offering a scalable and reliable solution for hydrological applications such as rainfall infiltration, groundwater recharge, and slope stability assessment.

The peak strength is a significant parameter in rock engineering, the traditional empirical strength criteria for rocks show good agreement with test results under specific conditions. However, it is not completely accurate for a wide range of loading stress domains and uncorrelated rock types. In this research, porosity, uniaxial compressive strength (UCS) and confining pressure are selected as input variables, and the artificial bee colony (ABC) algorithm is used to optimize the support vector machine (SVM) model. Finally, we validate and comparatively analyze the applicability of the models based on the testing set and the comprehensive evaluation indexes (namely correlation coefficient (R2), root mean square error (RMSE) and mean absolute percentage error (MAPE)). Meanwhile, the cosine amplitude method is applied to analyze the correlation between the peak strength and the input variables. The results indicate that both SVM model and ABC-SVM model are suitable for the prediction of peak strength under triaxial compression. Additionally, the ABC-SVM model obviously has better prediction performance by comparison.

The effect of freeze-thaw (F-T) cycles on the mechanical behaviors and internal mechanism of rock mass is a critical research topic. In permafrost or seasonally frozen regions, F-T cycles have adverse effects on the mechanical properties of rock mass, leading to many serious disasters in mining and geotechnical operations. In this paper, uniaxial compression tests are carried out on cyan sandstone after different F-T cycles. The failure modes and damage evolution of cyan sandstone under F-T cycles are studied. In addition, from the perspective of fracture and pore volume, the calculation equations of rock strain under frost heaving pressure and F-T cycles are established and verified with the corresponding laboratory tests. Subsequently, based on the classical damage theory, the F-T damage variables of cyan sandstone under different F-T cycles are calculated, and the meso-damage calculation model of cyan sandstone under F-T-loading coupling conditions is derived. Furthermore, through the discrete element numerical simulation software (PFC3D), the microscopic damage evolution process of cyan sandstone under uniaxial compression after F-T cycles is studied, including the change of microcracks number, distribution of microcracks, and the acoustic emission (AE) count. The goal of this study is to investigate the damage evolution mechanism of rock from the mesoscopic and microscopic aspects, which has certain guiding value for accurately understanding the damage characteristics of rock in cold regions.

Elastic modulus is a crucial mechanical parameter that measures the stiffness property of rock materials. In underground rock engineering, such as deep energy development and geological disposal of high-level nuclear waste, accurately determining the elastic modulus of engineering rock masses in high-temperature environments plays a vital role in understanding the instability-triggering conditions of hard and brittle rock masses in deep underground engineering and maintaining the safety and stability of deep underground engineering. Traditionally, conducting indoor experimental tests is time-consuming, labor-intensive, and costly. Therefore, developing a convenient and accurate rock elastic modulus prediction model based on machine learning technology holds significant importance. To this end, this study proposed a multi-step hybrid ensemble model (MHEM) for predicting high-temperature treated rock elastic modulus. The input parameters of the model include diameter, height, density, temperature, confining pressure, crack damage stress, and strength, while the output parameter is the elastic modulus. Subsequently, a coronavirus herd immunity optimizer (CHIO) intelligent optimization algorithm was employed to optimize the MHEM, and the CHIO-MHEM was established. Then, the performance of the developed models was compared and evaluated with eight other different prediction models. Finally, SHAP method and tree model feature analysis were used to quantify the feature importance of the prediction model. The research results indicate that, compared with other prediction models, the CHIO-MHEM exhibits superior performance, achieving accurate prediction of the elastic modulus of rocks treated at different high temperatures. Additionally, among all input parameters, rock density and temperature have the most significant influence on the rock elastic modulus.

Landslide dams, formed by natural disasters or human activities, pose significant challenges for lifespan prediction, which is crucial for effective water conservancy management and disaster prevention. This study proposes a hybrid CNN–Transformer model optimized using the Improved Black-Winged Kite Algorithm (IBKA) aimed at improving the accuracy of landslide dam lifespan prediction by combining local feature extraction with global dependency modeling. The model integrates CNN’s local feature extraction with Transformer’s global modeling capabilities, effectively capturing the nonlinear dynamics of key parameters affecting landslide dam lifespan. The IBKA ensures optimal parameter tuning, which enhances the model’s adaptability and generalization, especially when dealing with small-sample datasets. Experiments utilizing multi-source heterogeneous datasets compare the proposed model with traditional machine learning and deep-learning approaches, including LightGBM, MLP, SVR, CNN–Transformer, and BKA–CNN–Transformer. The results show that the IBKA–CNN–Transformer achieves R2 values of 0.99 on training data and 0.98 on testing data, surpassing the baseline methods. Moreover, SHapley Additive exPlanations analysis quantifies the influence of critical features such as dam length, reservoir capacity, and upstream catchment area on lifespan prediction, improving model interpretability. This approach not only provides scientific insights for risk assessment and decision making in landslide dam management but also demonstrates the potential of deep learning and optimization algorithms in broader geological disaster management applications.

Aiming at the problem of surrounding rock instability easily induced by high ground stress in the process of deep-well mining, the optimization of stope structure parameters is studied by combining numerical simulation with theoretical analysis. Firstly, the physical and mechanical properties of rock mass are fully understood using laboratory experiments. Then, six kinds of stope structure parameter schemes are preliminarily designed using the Matthews chart method. According to the geological conditions of the Ruihai Gold Mine, a large three-dimensional numerical model is established. Based on FLAC3D, the follow-filling continuous mining method is used to simulate the six schemes. By analyzing the influence and law of different stope structures on the stress, displacement, and plastic zone evolution of surrounding rock, the most effective mining strategy to balance the safety and economic benefits of the target area is determined. In the area with good rock mass quality, the optimal stope dimensions are 20 m in height, 15 m in width, and 80 m in length. In the rock mass area with fault crossing or relatively developed joint fissures, a reduced configuration of 20 m height, 10 m width, and 70 m length is recommended to enhance stability and stress management. Finally, comparative analysis of mining methods confirms that the follow-filling continuous mining method effectively mitigates ground pressure, offering a theoretical foundation for the safe and efficient extraction of deep mineral resources.

The accurate prediction of peak particle velocity (PPV) is essential for effectively managing blast-induced vibrations in mining operations. This study presents a novel PPV prediction method based on the social network search and LightGBM (SNS-LightGBM) deep gradient cooperative learning framework. The SNS algorithm enhances LightGBM’s learning process by optimizing hyperparameters through global search capabilities and balancing model complexity to improve generalization. To assess its performance, five baseline machine learning models and a hybrid model combining SNS-LightGBM were developed for comparison. The predictive performance of these models was evaluated using metrics such as coefficient of determination (R2), mean absolute error (MAE), mean absolute percentage error (MAPE), mean squared error (MSE), and root mean squared error (RMSE). The results indicate that the SNS-LightGBM model substantially improves both the accuracy and stability of PPV predictions. The SNS-LightGBM model outperformed all other models, achieving an R2 of 0.975, MAE of 0.086, MAPE of 0.071, MSE of 0.019, and RMSE of 0.138. Additionally, a feature importance analysis revealed that distance and charge weight are the most significant factors influencing PPV, far surpassing other parameters. These findings offer valuable insights for improving the precision of blast vibration prediction and optimizing blasting designs.

Accurate prediction of the longevity of Ldam, as temporary or permanent hydraulic structures, is crucial for ensuring downstream safety of lives and properties. This study proposes an intelligent forecasting model to accurately predict the longevity of Ldam. Firstly, a database containing a large amount of Ldam data is collected and organized, with key factors selected as indicators for the prediction system. Statistical indicators of the database are calculated, and in-depth analysis is conducted using correlation heatmaps and violin plots. Secondly, an intelligent forecasting model is constructed based on an improved intelligent optimization algorithm and ensemble learning. The model consists of multiple base learners (MLP, SVR, CatBoost) and a meta-learner (LightGBM). To effectively improve model performance, an improved intelligent optimization algorithm called IGTO is proposed to optimize the hyperparameters of the meta-learner within the stacked ensemble learning framework. In the analysis of the model's prediction results, prediction plots and regression plots are provided, and a series of evaluation metrics (R2, Adj-R2, RMSE, MAE, MAPE, VAF) are calculated. The results demonstrate that the IGTO stacked model exhibits high accuracy and reliability in predicting the longevity of Ldam, with computed values of R2 = 0.98, Adj-R2 = 0.98, RMSE = 93.98, MAE = 48.59, MAPE = 0.46, VAF = 98.29, showing good agreement with actual observed values. Furthermore, the model outperforms other prediction models and previous empirical formulas, validating the effectiveness and practicality of the IGTO improved optimization algorithm and ensemble learning framework. Additionally, the SHAP method is employed to assess the importance and impact of each input parameter on the model's predictions, quantifying the significance of each influencing factor on Ldam longevity and providing reference for engineering professionals. Moreover, collaborations were conducted with technical personnel from a local enterprise in Yangquan City, Shanxi Province, where on-site investigations were carried out and first-hand data of 46 sets of Ldam were obtained using drone-based 3D laser scanning technology. These data further enhance the value of this study, enrich the currently scarce database of Ldam, and further validate the generalization ability of the model and its effectiveness in future practical engineering applications. In conclusion, this study enriches the currently scarce database of Ldam and provides an effective method for accurate longevity prediction, demonstrating significant practical significance.

Rockburst is an extremely hazardous geological disaster. In order to accurately predict the hazardous degree of rockbursts, this paper proposes eight new classification models for predicting the intensity level of rockbursts based on intelligent optimisation algorithms and deep learning techniques and collects 287 sets of real rockburst data to form a sample database, in which six quantitative indicators are selected as feature parameters. In order to validate the effectiveness of the constructed eight machine learning prediction models, the study selected Accuracy, Precision, Recall and F1 Score to evaluate the prediction performance of each model. The results show that the NGO-CNN-BiGRU-Attention model has the best prediction performance, with an accuracy of 0.98. Subsequently, engineering validation of the model is carried out using eight sets of real rockburst data from Daxiangling Tunnel, and the results show that the model has a strong generalisation ability and can satisfy the relevant engineering applications. In addition, this paper also uses SHAP technology to quantify the impact of different factors on the rockburst intensity level and found that the elastic strain energy index and stress ratio have the greatest impact on the rockburst intensity level.

To quantify the disturbance range of the metal ore caving method, a new method to predict the disturbance range based on rock mechanics parameters is proposed. By combining the rock mechanics index with random medium theory, the prediction correction formula of the disturbance range is established. Based on disturbance equivalent centre and disturbance attenuation sphere, the two-dimensional safety criterion of mining disturbance in metal mines is derived. The new method is applied to 6 mines and the Jianshan Iron Mine, and the numerical simulation of the Jianshan Iron Mine by FLAC3D model is performed. The posteriori error ratio between predicted value and measured value is 0.0357, and the multi-factor cross analysis shows that the relative error of this method to the result is only 0.078. It is found that the method has high prediction accuracy and does not significantly affect the security.

The mechanical characteristics of rock are greatly influenced by hydrochemical corrosion. The chemical corrosion impact and deformation properties of the meso-pore structure of rock under the action of different hydrochemical solutions for the stability evaluation of rock mass engineering are of high theoretical relevance and applied value. Based on actual data, a support vector machine (SVM) rock constitutive model based on artificial bee colony algorithm (ABC) optimization is constructed in this article. The impact of porosity (chemical deterioration), confining pressure, and other aspects is thoroughly examined. It is used to mimic the triaxial mechanical behavior of rock under various hydration conditions, with high nonlinear prediction ability. Simultaneously, the statistical damage constitutive model and the ABC-SVM constitutive model are used to forecast the sample’s stress–strain curve and compare it to the experimental data. The two models’ correlation coefficients (R2), root mean square error (RMSE), and mean absolute percentage error (MAPE) are computed and examined. The correlation coefficient between the ABC-SVM constitutive model calculation results and the experimental results is found to be larger (R2 = 0.998), and the error is smaller (RMSE = 0.7730, MAPE = 1.51), indicating that it has better prediction performance on the conventional triaxial constitutive relationship of rock. It is a highly promising new way of describing the rock’s constitutive connection.

A large number of rock works in cold areas suffer from long-term freeze-thaw damage, and it seriously affects the stability of mine slopes. In this paper, the XRD component measurement, P-wave velocity, freeze-thaw cycling test at different times, uniaxial compression test, and scanning electron microscope (SEM) test were carried out to obtain the mechanical properties and microstructure evolution of skarn under the effect of freeze-thaw cycles. The results of the study indicate that with an increase in the number of freeze-thaw cycles, the mass of the rock gradually increases and the P-wave velocity, uniaxial compressive strength, elastic modulus, and Poisson's ratio all decrease. Based on the SEM image of the rock after crushing, fine pores and fissures gradually developed, expanded, and penetrated each other under the action of freezing and thawing; the inter-particle bonding force decreased; and the cement gradually loosened. The fractal dimension of the specimens under different numbers of freeze-thaw cycles was obtained using the box dimension method, and the degradation of the fine structure of the rock was quantitatively elaborated. By establishing the relationship between the compressive strength of rocks and the fractal dimension, the mechanism of damage to skarn under freeze-thaw action was further investigated. It provides some theoretical basis for the characterization of freeze-thaw damage of rocks in cold regions.

In the drifts of underground metal mines, the extraction of rock mass discontinuity characteristics from point cloud models generated with laser scanning has become the main approach. However, the exposure of discontinuities is restricted in drifts, and the size of discontinuities cannot be measured directly. Therefore, it is necessary to use a reasonable sampling tool to estimate the mean trace length of the discontinuities that are mapped in the point cloud model. In this paper, a method to estimate the mean trace length of discontinuities using a three-dimensional (3D) model of a drift (3DM) is proposed. Through the point cloud data of a drift obtained using 3D laser scanning, the information on discontinuities in the surrounding rock was extracted; then, the mean trace length was estimated using 3DEC to set sampling windows on the roof and sidewall in the 3DM. By analyzing the difference between the circular sampling window and the rectangular sampling window using simulated cases, the estimation results showed that the mean trace length obtained using circular measuring windows in the 3DM was closer to the true trace length. Finally, the method was used in a practical engineering case in Jianshan Iron Mine, Panzhihua, Sichuan, China.