The occurrence of rockburst can cause significant disasters in underground rock engineering. It is crucial to predict and prevent rockburst in deep tunnels and mines. In this paper, the deficiencies of ensemble learning algorithms in rockburst prediction were investigated. Aiming at these shortages, a novel machine learning model, deep forest, was proposed to predict rockburst risk. The deep forest combines the characteristics of deep learning and ensemble models, which can solve complex problems. To develop the deep forest model for rockburst prediction, 329 real rockburst cases were collected to build a comprehensive database for intelligent analysis. Bayesian optimization was proposed to tune the hyperparameters of the deep forest. As a result, the deep forest model achieved 100% training accuracy and 92.4% testing accuracy, and it has more outstanding capability to forecast rockburst disasters compared to other widely used models (i.e., random forest, boosting tree models, neural network, support vector machine, etc.). The results of sensitivity analysis revealed the impact of variables on rockburst levels and the applicability of deep forest with a few input parameters. Eventually, real cases of rockburst in two gold mines, China, were used for validation purposes while the needed data sets were prepared by field observations and laboratory tests. The promoting results of the developed model during the validation phase confirm that it can be used with a high level of accuracy by practicing engineers for predicting rockburst occurrences.

The semi-circular bend (SCB) specimen is widely used to measure fracture toughness of brittle materials such as rock. In this work, the stress field, fracture process zone (FPZ), and crack propagation velocity of SCB specimen are analyzed during the fracture process of rock specimens. The FPZ of specimen is obtained by experimental and numerical methods under a three-point bend test. The stress concentration zones of specimen present a heart shape at peak load points. FPZ forms before macro fracture occurs. The macro fracture form inside FPZ in a post-peak region of a load–displacement curve. The crack propagation process of specimen include two stages, namely the rapid crack initial development stage, and the final crack splitting stage. The maximum crack propagation velocity of specimen is about 267 m/s, and the average crack propagation velocity is about 111 m/s.

The Brazilian disc test is a popular tensile strength test method for engineering materials. The fracture behavior of specimens in the Brazilian disc test is closely related to the validity of the test results. In this paper, the fracture process of granite discs under different loading configurations is simulated by using a coupled finite–discrete element method. The results show that the maximum tensile stress value is located within 18 mm (0.7 times the disc radius) of the vertical range of the disc center under different loading configurations. In small diameter rods loading, the invalid tensile strength is obtained because the crack initiation and plastic strain is at the end of the disc. The crack initiation points of flat platen loading and curved jaws loading are all within the center of the disc, and the valid tensile strength can be obtained. The tensile strength test results under different loading configurations show that the error of small diameter rods loading is 13%, while the errors of flat platen loading and curved jaws loading are both 1%. The curved jaws loading is the most suitable for measuring the tensile strength of brittle materials such as rock, followed by flat platen loading. The small diameter rods loading is not recommended for the Brazilian test.

The occurrence of rockburst can seriously impact the construction and production of deep underground engineering. To prevent rockburst, machine learning (ML) models have been widely employed to predict rockburst based on some related variables. However, due to the costs and complicated geological conditions, complete datasets to evaluate rockburst cannot always be obtained in rock engineering. To fill this limitation, this study proposed an ensemble tree model suitable for incomplete datasets, i.e., the histogram gradient boosting tree (HGBT), to build intelligent models for rockburst prediction. Three hundred fourteen rockburst cases were employed to develop the HGBT model. The hunger game search (HGS) algorithm was implemented to optimize the HGBT model. The established HGBT model had an excellent testing performance (accuracy of 88.9%). An incomplete database with missing values was applied to compare the performances of HGBT and other ML models (random forest, artificial neural network, and so on). HGBT received an accuracy of 78.8% in the incomplete database, and its capacity was better than that of other ML models. Additionally, the importance of input variables in the HGBT model was analyzed. Finally, the feasibility of the HGBT model was validated by rockburst cases from Sanshandao Gold Mine, China.

Deep rock engineering is in a high pre-stressed state before excavation. In this research, a method to calculate the release of strain energy caused by excavation in pre-stressed rock is proposed. The normal stress release after excavation leads to a reduction in strain energy in rock specimens. The influence of excavation height and width on strain energy release is inconsistent under vertical loading. When the height of the hole is 1 mm, the strain energy release is large, and the increase in height of hole leads to a slow increase in the strain energy release. When the width of the hole is 1 mm, the strain energy release is very small, and the increase in the width of the hole leads to an increasingly faster release of strain energy. This strain energy release exponentially increases with the increase in the lateral pressure coefficient, showing a trend in the second power of the lateral pressure coefficient. Moreover, the tunnel failure caused by excavation under high stress is obtained by a numerical calculation. The failure modes of the deep tunnel model are strain rockbursts caused by tangential stress concentrations and spalling caused by normal stress release, which is also observed in the failure mode of the actual tunnel.

The evaluation of rockburst damage potential plays a significant role in managing rockburst risk and guaranteeing the safety of personnel. However, it is still a challenging problem because of its complex mechanisms and numerous influencing factors. In this study, a bagged ensemble of Gaussian process classifiers (GPCs) is proposed to assess rockburst damage potential with an imbalanced dataset. First, a rockburst dataset including seven indicators and four levels is collected. To address classification problems with an imbalanced dataset, a novel model that integrates the under-sampling technique, Gaussian process classifier (GPC) and bagging method is constructed. Afterwards, the comprehensive performance of the proposed model is evaluated using the values of accuracy, precision, recall, and F₁. Finally, the methodology is applied to assess rockburst damage potential in the Perseverance nickel mine. Results show that the performance of the proposed bagged ensemble of GPCs is acceptable, and the integration of data preprocessing, under-sampling technique, GPC, and bagging method can improve the model performance. The proposed methodology can provide an effective reference for the risk management of rockburst.