Abstract:Microseismic monitoring technique has been proved to be an effective method to predict catastrophic disasters in underground engineering. However, it is difficult to identify the microseismic signals of coal fracture due to the complex environment and operations in coal mine, which hinders the application of microseismic monitoring on the early warning of coal and gas outbursts (outbursts). In this work, microseismic signals in the heading face of Xinyuan coal mine were recorded under different operations, including drilling of holes, roadway support, moving of belt conveyor, driving activities, blasting and no work. The records show that the microseismic events occur frequently under driving activities, while a few microseismic events happen during the no work period. The characteristics of time domain, frequency domain and time–frequency for different microseismic signals were then analyzed based on the fast Fourier transform and short-time Fourier transform methods. The results indicate that the low-frequency waveform is well developed under the drilling of holes, roadway support, moving of belt conveyor and no work period. A short and discontinuous high-frequency waveform exists in the work of roadway support, moving of belt conveyor and no work period, which belongs to the coal fracture microseismic signals. Driving activities have massive low-frequency and high-frequency waveforms, but the high-frequency waveform corresponds to the coal fracture microseismic signals. The high-frequency waveform existing in the blasting work presents an obvious periodic feature. Combined with the long short-term memory (LSTM) network, different MS signals were further classified after extracting the maximum amplitude, dominant frequency, rising time, duration time and period from separation waveforms. Compared with the k-nearest neighbor and support vector machine models, the classification result of LSTM has the lowest loss value and highest accuracy rate of about 0.18 and 85%. The microseismic events of coal fracture identified by the LSTM have a better sequential response to gas concentration, which can be used to warn the outbursts in advance. This study provides a theoretical support for the early warning of outbursts and is meaningful for the safety and efficient production of coal mines.

Response Characteristics of Gas Concentration Level in Mining Process and Intelligent Recognition Method Based on BI-LSTM

Research on coal mine longwall face gas state analysis and safety warning strategy based on multi-sensor forecasting models

Intelligent Prediction of Coal Mine Gas Emission Based on Deep Neural Network

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

Gas Outburst Warning Method in Driving Faces: Enhanced Methodology through Optuna Optimization, Adaptive Normalization, and Transformer Framework

Method and Validation of Coal Mine Gas Concentration Prediction by Integrating PSO Algorithm and LSTM Network

LSTM-Pearson Gas Concentration Prediction Model Feature Selection and Its Applications

Predicting Methane Concentrations in Underground Coal Mining Using a Multi-Layer Perceptron Neural Network Based on Mine Gas Monitoring Data

Projection of Early Warning Identification of Hazardous Sources of Gas Explosion Accidents in Coal Mines Based on NTM Deep Learning Network

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

Combined Prediction Model of Gas Concentration Based on Indicators Dynamic Optimization and Bi-LSTMs

Multiple Sequence Long and Short Memory Network Model for Corner Gas Concentration Prediction on Coal Mine Workings

Research on Gas Concentration Prediction Based on the ARIMA-LSTM Combination Model

Monitoring and Control Model for Coal Mine Gas and Coal Dust

Distributed gas concentration prediction with intelligent edge devices in coal mine

Sound Recognition Method of Coal Mine Gas and Coal Dust Explosion Based on GoogLeNet

Gas concentration prediction by LSTM network combined with wavelet thresholding denoising and phase space reconstruction

Methane Concentration Prediction Method Based on Deep Learning and Classical Time Series Analysis

Time Series Prediction of Gas Emission in Coal Mining Face Based on Optimized Variational Mode Decomposition and SSA-LSTM

Early Warning of Gas Concentration in Coal Mines Production Based on Probability Density Machine

Prediction of Gas Concentration Based on LSTM-LightGBM Variable Weight Combination Model