In order to accurately predict the gas concentration, find out the gas abnormal emission in advance, and take effective measures to reduce the gas concentration in time, this paper analyzes multivariate monitoring data and proposes a new dynamic combined prediction method of gas concentration. Spearman’s rank correlation coefficient is applied for the dynamic optimization of prediction indicators. The time series and spatial topology features of the optimized indicators are extracted and input into the combined prediction model of gas concentration based on indicators dynamic optimization and Bi-LSTMs (Bi-directional Long Short-term Memory), which can predict the gas concentration for the next 30 min. The results show that the other gas concentration, temperature, and humidity indicators are strongly correlated with the gas concentration to be predicted, and Spearman’s rank correlation coefficient is up to 0.92 at most. The average R² of predicted value and real value is 0.965, and the average prediction efficiency R for gas abnormal or normal emission is 79.9%. Compared with the other models, the proposed dynamic optimized indicators combined model is more accurate, and the missing alarm of gas abnormal emission is significantly alleviated, which greatly improves the early alarming accuracy. It can assist the safety monitoring personnel in decision making and has certain significance to improve the safety production efficiency of coal mines.

In order to realize dynamic, continuous, and real-time prediction of coal and gas outburst risk in real time in blasting driving face, an outburst risk prediction method based on the characteristics of gas emission after blasting is proposed. In this study, the causes of abnormal gas concentration in blasting driving face are analyzed, and the identification method of abnormal gas concentration based on weighted K-nearest neighbor (weighted KNN) is proposed. The correlation between gas emission characteristics after blasting and K₁ value is analyzed, and the prediction model of outburst risk based on convolutional neural networks (CNN) is established and applied in Jinjia coal mine in China. The results show that the causes of abnormal gas concentration mainly include ventilation stop, blasting operation, sensor adjustment, and other abnormalities. The accuracy of the identification method is 86.1. Especially, the identification accuracy of blasting operation is 92. There are strong correlations between the growth rate, peak value, and decay rate of gas concentration after blasting and K₁ value, and the maximum correlation coefficient is 0.92. Using the prediction model, 28 times of jet holes and 1 small outburst event are predicted successfully, and the efficiency of the prediction model is 76.39. By this technology, the utilization rate of gas information is improved, and the relationship between the change characteristics of gas concentration after blasting and the risk of coal seam outburst is established, which is of significant for improving the prediction accuracy and risk management ability of coal and gas outburst.

The growing burial depth increases the potential risk of ground stress-induced coal and rock dynamic disasters in coal mining. This study attempts to use microseismic and computed tomography (MS-CT) to quantitatively characterize the stress field in coal seam areas. A comprehensive approach encompassing laboratory experiments, numerical simulations, and field investigations was employed to accomplish this objective. The experimental findings suggest a power function relationship between the wave velocity of coal and stress. The model relating wave velocity ratio to stress proves more suitable for research on stress field quantification. This model captures the positive correlation between wave velocity and stress changes and eliminates the influence of sample differences through normalization, rendering the fitting parameter b applicable in the field. This paper takes the No.22 coal seam of Jinjia coal mine as the research object; MS-CT technology was used to generate the cloud map of wave velocity field distribution in the 11224 working face and combined with the experimentally established relational model to relate the wave velocity ratio to the stress and calculate the results of quantitative characterization of the regional stress. At the same time, a detailed three-dimensional numerical model was established and simulated to obtain the stress distribution in the 11224 working face area. Comparison of the simulated stress field results with the quantitative results of MS-CT shows that the general trend is basically the same, and the stress is higher in the area with greater burial depth. The area affected by the overlying goaf has lower stress in the coal seam. The maximum error value of the two is within ± 2 MPa, and the correlation coefficient is 0.79, indicating a strong correlation. Therefore, the MS-CT technique combined with the experimental wave velocity-stress coupling relationship model can determine the stress field distribution and realize the quantitative stress characterization. This provides a crucial foundation for analyzing the mechanical model of coal seam instability and calculating the judgment index of coal-rock dynamic disaster excitation.

Coal and gas outburst poses a significant threat to mine safety, requiring effective risk level prediction methods. To enhance the accuracy and intelligence of outburst risk prediction, a prediction indices system of outburst risk integrating both prior and current information is constructed, the outburst risk level classification method based on Bayesian–Gaussian mixture model is proposed, the outburst risk level prediction model based on Bayesian neural network is established. The proposed methodology is validated using field data from an engineering site. The results show that the outburst risk can be classified into four risk levels: level I, level II, level III and level IV. The model achieves prediction accuracies of 97.94%, 97.26%, 96.36%, and 90.91% for different time windows (1 h, 2 h, 4 h, and 8 h, respectively). The 2 h time window is identified as optimal, ensuring over 90% accuracy across all risk levels, including 100% for level IV. Furthermore, the model successfully predicts abnormal events such as drilling anomalies, abnormal effect test, and gas emission at the Ji₁₅-15 080 machine roadway in No.8 Coal Mine. Beyond risk level prediction, the model quantifies the probability of each risk level, providing a more interpretable decision-making for mine safety management. The research contributes to coal and gas outburst disaster prevention and provide valuable insights for enhancing mining safety.