An accurate assessment of the stress concentration of the surrounding rock is crucial for ensuring safety in underground coal mines with a high potential for rockburst accidents. Traditional methods for measuring the stress of the surrounding rock use borehole stress-monitoring equipment and drilling and cutting methods. However, these methods do not always yield accurate and reliable data. In this study, we aim to explore the feasibility of a novel approach for characterizing stress concentration in surrounding rock using monitoring while drilling technology during large-diameter drilling. This study combined microseismic monitoring to collect vibration signals during the drilling process and collection of pulverized coal samples at each drilling stage. We analyzed the amplitude, number of coal vibration events and proportion of coarse pulverized coal. By integrating these indices, we characterized the stress concentration in the surrounding rock. Our approach was validated by comparing accurate, stable and representative data from deep- and shallow-hole stress gauges installed in similar locations and data from the conventional drilling cutting method. Our findings indicate that the proposed method provides a reliable and effective alternative to traditional techniques during large-diameter drilling. The proposed approach can significantly enhance safety management in underground coal mines prone to rockburst accidents.

Microseismic monitoring is a critical technology for mine safety monitoring, but existing microseismic picking methods exhibit instability and limitations when dealing with high-noise data in coal mine environments. This paper proposes a method for improving the picking of high-noise microseismic P-wave first arrivals, called IWTSE-MSDACF-AIC. The method first uses the improved complete ensemble empirical mode decomposition with adaptive noise to decompose the microseismic signal into a series of intrinsic mode functions (IMFs). Then, the sample entropy of the IMFs is calculated, and an appropriate threshold is set to perform wavelet denoising on the IMFs. The signals are then reconstructed to distinguish noise from useful signals. Finally, the denoised signal’s P-wave first arrival is automatically determined using the proposed picking method based on the moving standard deviation-adaptive characteristic function and Akaike information criterion, which incorporates the relative energy coefficient and relative energy time series. Tests using synthetic seismic records with different signal-to-noise ratios and validation on real coal mine seismic datasets show that the proposed denoising strategy and picking method achieve high accuracy and robustness. In practical data tests, 90.01% of data errors fell within the range of 0s to 0.06s, demonstrating excellent picking performance. Furthermore, in grid search localization using five calibration blasts at the Dongtan coal mine, the localization results based on the proposed method significantly outperformed those based on traditional methods and PhaseNet.

Accurate temporal resolution of the eikonal equation forms the cornerstone of seismological studies, including microseismic source localization, and travel-time tomography. Physics-informed neural networks (PINNs) have gained significant attention as an efficient approximation technique for numerical computations. In this study, we put forth a novel model named Eiko-PIGCNet, a graph convolutional neural network that incorporates physical constraints. We demonstrate the effectiveness of our proposed model in solving the 3D eikonal equation for travel-time estimation. In our approach, the discretized grid points are converted into a graph data structure, where every grid point is regarded as a node, and the neighboring nodes are interconnected via edges. The node characteristics are defined by incorporating the velocity and spatial coordinates of the respective grid points. Ultimately, the efficacy of the Eiko-PIGCNet and PINNs is evaluated and compared under various velocity models. The results reveal that Eiko-PIGCNet outshines PINNs in terms of solution accuracy and computational efficiency.

Microseismic monitoring is pivotal for mine safety, offering early warnings for coal mining operations. By introducing the MTfit and MSATSI inversion methods to study the focal mechanisms of microseismic events and the stress field in coal mines, we applied these techniques for the first time to the 63_upper06 working face in the Dongtan Coal Mine. We selected 107 high-quality microseismic events during the mining process and used MTfit to invert their focal mechanisms. Moreover, we used MSATSI to determine the direction of the maximum horizontal principal compressive stress in the region. This allowed us to establish a connection between mining activities, focal mechanisms, and rock strata fractures. Results show the predominance of normal and reverse faulting with significant CLVD components, underlining tensile and compressive failures as dominant in the mine’s seismicity. Further, the study subdivides the research area into three subregions for detailed stress analysis: Subregion 1 is located below the working face with a maximum principal stress of 20.3 MPa, oriented between N60°E and N75°E; Subregion 2 is above the working face and exhibits more complex stress conditions with a maximum principal stress of 30.1 MPa; and Subregion 3 is situated at the roof’s right side of the working face, with a maximum principal stress of 20.3 MPa, oriented at 30°northwest. It provides an important reference for seismic risk assessment and disaster prevention and control during mining.