Abstract:A thorough excavation of the deep ultra-thick mines is highly challenging in rock engineering, which mainly depends on the rheological properties of coal-rock mass, high-stress concentrations, and complex geological conditions. Under heterogeneous settings, a detailed understanding of microseismic (MS) spatio-temporal response to rock mechanical behavior is essential for the efficient and safe-yield of deep ultra-thick mines. This paper utilizes real-time passive seismological data of 36 months in an ultra-thick coal mine through LLTCC mining to quantify the coal-rock mass mechanical behavior. In the post-processing of geophysical data, Power Spectral Density (PSD) was performed and the MS source parameters were computed containing over 18,000 events. The events were classified into different energy-levels to evaluate the increased-pressure rock mass. The Gaussian results reveal that the distribution of MS events coincides with the increased rock pressure. The high energy events (≥ 50,000 J) accounted for 57.93% of the microseismic-induced earthquakes. Moreover, the cloud energy density maps were obtained to identify the zones of high-stress concentration and enhanced seismicity. It is inferred that the overburden pressure and intense fracturing developed the high-stress concentration zones during excavation. Furthermore, the temporal parameters computed from MS data show that sharp-rise and sharp-drop variation can be regarded as early warning indicators for increased rock pressure. High deviation in energy and frequency also evidenced the increased rock pressure and micro-fractures to macro-fracture development. Based on these precursory parameters, a comprehensive early warning method was proposed. Besides, 3D visualization of the fracturing process and identification of new faults assisted in the accurate evaluation of the ultra-thick mine. Several risk zones were identified based on the seismic energy and fractures correlation. Finally, with the help of a deep learning LSTM approach, a prominent peak in time-series data is predicted that defines the increased rock pressure and further characterizes the coal-rock mechanical response. Compared with the existing studies, our integrated geophysical and mine engineering workflow can provide accurate insights to depict the rock mass mechanical behavior of the deep mines as an aid for disaster prediction and early warning.

LGF SeismoLocator: A Deep Learning Model for Precision Microseismic Event Localization in Coal Mines

Improved Mine Microseismic Event Recognition Method Combining Neural Network and Transfer Learning.

Microseismic Event Recognition and Transfer Learning Based on Convolutional Neural Network and Attention Mechanisms

Intelligent Location of Microseismic Events Based on a Fully Convolutional Neural Network (FCNN)

A novel identification method of microseismic events based on empirical mode decomposition and artificial neural network features

Characteristics and Classification of Microseismic Signals in Heading Face of Coal Mine: Implication for Coal and Gas Outburst Warning

Extracting and Predicting Rock Mechanical Behavior Based on Microseismic Spatio-temporal Response in an Ultra-thick Coal Seam Mine

Coal-Rock Image Recognition Method for Complex and Harsh Environment in Coal Mine Using Deep Learning Models

Large-Scale Underground Mine Positioning and Mapping with LiDAR-Based Semantic Intersection Detection

The method for precise seismic detection of geological structures in underground coal mines and application

A LiDAR SLAM Algorithm Considering Dynamic Extraction of Feature Points in Underground Coal Mine

ActiveGCN-CT: Active Learning-Enhanced Graph Convolutional Networks for Computed Tomography Inversion in Coal Mines

High-Precision Microseismic Source Localization Using a Fusion Network Combining Convolutional Neural Network and Transformer

Automatic Classification of Microseismic Signals Based on MFCC and GMM-HMM in Underground Mines

Research on coal-rock identification method and data augmentation algorithm of comprehensive working face based on FL-Segformer

Microseismic Location in Hardrock Metal Mines by Machine Learning Models Based on Hyperparameter Optimization Using Bayesian Optimizer

Deep learning for low-magnitude earthquake detection on a multi-level sensor network

Securing Fast and High-Precision Localization for Shallow Underground Explosive Source: A Curiosity-Driven Deep Reinforcement Learning Approach

Coalbed methane concentration prediction and early-warning in fully mechanized mining face based on deep learning

Detection and Monitoring of Mining-Induced Seismicity Based on Machine Learning and Template Matching: A Case Study from Dongchuan Copper Mine, China

Fine Classification Method for Massive Microseismic Signals Based on Short-Time Fourier Transform and Deep Learning