This paper proposes a new method for predicting slope safety factors that combines convolutional neural networks (CNNs), gated recurrent units (GRUs), and attention mechanisms. This method can better capture long-term dependencies, enhance the ability to model sequential data, and reduce the dependence on noisy data, thereby reducing the risk of overfitting. The goal is to improve the accuracy of slope safety factor prediction, detect potential slope stability issues in a timely manner, and take corresponding preventive and control measures to ensure the long-term stability and safety of infrastructure and promote sustainable development. The Pearson correlation coefficient is used to analyze the relationship between the target safety factor and the collected parameters. A one-dimensional CNN layer is used to extract high-dimensional features from the input data, and then a GRU layer is used to capture the correlation between parameters in the sequence. Finally, an attention mechanism is introduced to optimize the weights of the GRU output, enhance the influence of key information, and optimize the overall prediction model. The performance of the proposed model is evaluated using metrics such as the mean absolute error (MAE), mean absolute percentage error (MAPE), mean squared error (MSE), root-mean-square error (RMSE), and R². The results show that the CNN-GRU-SE model outperforms the GRU, CNN, and CNN-GRU models in terms of prediction accuracy for slope safety factors, with improvements of 4%, 2%, and 1%, respectively. Overall, the research in this paper makes valuable contributions to the field of slope safety factor prediction, and the proposed method also has the potential to be extended to other time-series prediction fields, providing support for a wide range of engineering applications and further promoting the realization of sustainable development.

The vibration of rock breaking in tunnel excavation may cause serious damage to nearby buildings if it is not controlled properly. With reference to a hard rock tunnel in China, the vibration response to the high-pressure gas expansion method (HPGEM), an emerging rock-breaking approach, was investigated with field tests, theoretical derivations, and numerical simulations, then comparisons with the traditional dynamite blast were performed. Firstly, the vibration velocity prediction formulas of the two methods were fitted based on the field tests. Subsequently, the accuracy of the formula was verified by numerical simulation, and the vibration attenuation law of the HPGEM was explored. Comparisons were made between the blast and HPGEM, particularly the differences in peak particle velocity (PPV) for different agent qualities, distance from the blasting center, and engineering conditions. Furthermore, this study also analyzed the relationship between the agent qualities and the rock-breaking volume under different cases, finding that the HPGEM has slight vibration and good rock-breaking effect. The HPGEM is thus fully capable of replacing dynamite blasting to carry out rock-breaking operations in certain special areas.

A precise rock mass grade result is crucial for directing the tunnel excavation engineering design. A novel rock mass classification method for tuff tunnel based on the high-pressure gas expansion method (HPGEM) was proposed, which was primarily built on field test data previously acquired by the research team. The main achievements are as follows: combined with field data and the HPGEM rock-breaking theory, analyzing the rock uniaxial compressive strength, rock mass integrity index, and the relationship between the gas generator unit consumption and fitted the relevant equations. After that, the rock optimal uniaxial compressive strength (about 150 MPa) and the rock integrity factor (about 0.85) were obtained. With reference to the BQ rock mass classification method, a new rock mass classification method that applied to HPGEM was proposed. This study fills the gap of the appropriate rock mass classification method requested on HPGEM.

A reasonable support parameter design scheme is the key to ensuring the stability of the roadway. This study established 217 models using FLAC3D to analyze the evolution characteristics of elastic strain energy and plastic dissipation energy of surrounding rock under different shotcrete and rockbolt support structures. Additionally, five single models (BP, DT, ELM, RF, SVM) were introduced to explore the application of machine learning in predicting the stability of the roadway. The study found that in the parameters of the shotcrete layer support structure, the energy evolution of the surrounding rock is more sensitive to isotropic and thickness; in the parameters of the anchor rod support structure, the energy evolution of the surrounding rock is more sensitive to Young’s modulus, cross-sectional area young, and grout stiffness. Additionally, the parameters of the shotcrete layer support structure are not necessarily the larger the better. When isotropic is 100 GPa, both the dissipated energy and the elastic strain energy are higher than that at 25 GPa. The results of the single model test indicate that machine learning is relatively accurate in predicting different shotcrete and anchor support structures. The runtime difference between traditional methods and machine learning models highlights the potential advantages of machine learning.