Predicting pillar stability is a vital task in hard rock mines as pillar instability can cause large-scale collapse hazards. However, it is challenging because the pillar stability is affected by many factors. With the accumulation of pillar stability cases, machine learning (ML) has shown great potential to predict pillar stability. This study aims to predict hard rock pillar stability using gradient boosting decision tree (GBDT), extreme gradient boosting (XGBoost), and light gradient boosting machine (LightGBM) algorithms. First, 236 cases with five indicators were collected from seven hard rock mines. Afterwards, the hyperparameters of each model were tuned using a five-fold cross validation (CV) approach. Based on the optimal hyperparameters configuration, prediction models were constructed using training set (70% of the data). Finally, the test set (30% of the data) was adopted to evaluate the performance of each model. The precision, recall, and F₁ indexes were utilized to analyze prediction results of each level, and the accuracy and their macro average values were used to assess the overall prediction performance. Based on the sensitivity analysis of indicators, the relative importance of each indicator was obtained. In addition, the safety factor approach and other ML algorithms were adopted as comparisons. The results showed that GBDT, XGBoost, and LightGBM algorithms achieved a better comprehensive performance, and their prediction accuracies were 0.8310, 0.8310, and 0.8169, respectively. The average pillar stress and ratio of pillar width to pillar height had the most important influences on prediction results. The proposed methodology can provide a reliable reference for pillar design and stability risk management.

Due to various environmental issues caused by resource exploitation, establishing green mines is an essential measure to realize sustainable growth for mining companies. This research aimed to develop a novel methodology to evaluate the performance of green mines within hesitant fuzzy conditions. First, hesitant fuzzy sets (HFSs) were used to express original fuzzy assessment values. Then, the extended expert grading approach and the modified maximum deviation method with HFNs were combined to determine comprehensive importance degrees of criteria. Afterward, the traditional qualitative flexible (QUALIFLEX) method was integrated with the Organísation, rangement et synthèse de données relationnelles (ORESTE) model to achieve the rankings of mines. Finally, the proposed hesitant fuzzy ORESTE–QUALIFLEX approach was utilized to evaluate the performance of green mines. In addition, the robustness of the method was verified by a sensitivity analysis, while the effectiveness and strengths were certified by a comparison analysis. The results indicate that the proposed methodology has great robustness and advantages and that it is feasible and effective for the performance evaluation of green mines under hesitant fuzzy environment.

As a powerful numerical analysis tool, PFC (Particle Flow Code) is widely applied to investigate the mechanical behavior of rock specimen or rock engineering under different stress states. To match the macroscopic properties of the PFC model with those of the rock, a set of micro-scale parameters of the model needs to be calibrated. Thus, this paper proposed an optimization method combining Box–Behnken experimental design and desirability function approach to quickly and accurately find the values of the micro-scale parameters. The sensitivity of the main micro-scale parameters (mean value of parallel-bond normal strength σ_c, ratio of particle normal to shear stiffness E_c, and Young’s modulus at each particle–particle contact k_n/k_s) and their interactions to the macroscopic responses (uniaxial compressive strength, Young’s modulus, and Poisson’s ratio) were thoroughly analyzed using response surface theory. After that, validation study was conducted on the calibrated model. The results manifest that the uniaxial compressive strength is extremely significantly affected by σ_c and k_n/k_s, the Young’s modulus is highly correlated with E_c and k_n/k_s, and the Poisson’s ratio is most significantly influenced by k_n/k_s. Additionally, the interaction of micro-scale parameters also has different impact upon the responses. Moreover, the simulated crack behavior around differently shaped openings in rock samples under uniaxial compression is found to be well agreeable with the experimental results, which verifies the reliability of the proposed method.