REAL-TIME SEISMIC WAVE VELOCITY DETERMINATION FOR ACCURATE EVENT SOURCE LOCATIONS IN UNDERGROUND MINES

Abstract

This study addresses the considerable impact of human-induced seismic activity in underground mining on safety, productivity, and operational costs. Accurate prediction of microseismic and rockburst source locations is crucial for mitigating unforeseen events such as rockbursts. Current limitations in predicting the timing of these events necessitate a focus on probable locations. The research aims to overcome this limitation by developing a methodology that tracks and adapts to changes in ground conditions, providing a real-time solution for selecting appropriate velocity models in seismic monitoring systems. By considering the constantly changing velocity model in underground mining environments, the approach seeks to improve the reliability of source location calculations, ultimately enhancing worker safety and mining productivity. The study recognizes the evolving characteristics of rock masses and voids during mining, highlighting the inadequacy of a constant velocity model in source location algorithms. Laboratory experiments, simulating underground mine conditions with continuous changes, generated data for this study. Analysis reveals that, due to the heterogeneity and continuous changes in the mining environment, seismic wave velocity cannot be treated as a constant in source location algorithms. Predicting real-time seismic wave velocities significantly enhances the accuracy of seismic event source location. Dynamic numerical modeling in FLAC3D, using laboratory data, was employed to understand wave propagation and the physics of the problem. Machine learning methodologies, including Linear Regression models trained on laboratory data and Deep Artificial Neural Networks for heightened precision, were harnessed to predict seismic wave velocities under varying conditions.