Machine learning-based microseismic catalog and passive seismic tomography evaluating the effect of grouting in Zhangji coal mine, China

Fault grouting is important in preventing groundwater inrush into coal mines from aquifers underlying the coal seams. However, evaluating the effect of grouting in the coal mine is difficult and expensive; hence, the region is sparsely explored. Therefore, we propose a more economical and efficient method for investigating grouting in the coal mine, which utilizes a real-time microseismic monitoring system. A system is an integrated approach involving a deep neural network phase picker method, grid search location method, and double-difference seismic tomography method for enhanced real-time processing of the microseismic data of 1613A working face in Zhangji coal mine, China. The velocity structure of the 1613A working face has been inverted using the microseismic events in September 2020. We found that the grouting treatment area is considerably perturbed during mining, with the 50-m area below the mining floor characterized by a high seismic velocity of up to 4.3 km/ s, showing optimal grouting. The checkerboard resolution test shows that the seismic velocity results in concern have high reliability. The effect of grouting is well evaluated by machine learning-based microseismic catalog and passive seismic tomography.