Title: Experimental study on prediction of gas pressure variation during coal and gas outburst

Authors: Erhui Zhang; Xukai Dong; Baokun Zhou; Lei Yang

Addresses: School of Emergency Management and Safety Engineering, China University of Mining and Technology Beijing, Ding 11, Xueyuan Road, Haidian District, Beijing 100083, China ' School of Energy and Mining Engineering, China University of Mining and Technology Beijing, Ding 11, Xueyuan Road, Haidian District, Beijing 100083, China ' School of Energy and Mining Engineering, China University of Mining and Technology Beijing, Ding 11, Xueyuan Road, Haidian District, Beijing 100083, China ' School of Energy and Mining Engineering, China University of Mining and Technology Beijing, Ding 11, Xueyuan Road, Haidian District, Beijing 100083, China

Abstract: To accurately predict the variation of gas pressure during the coal and gas outburst experiment, a gas pressure prediction model was established based on Keras and long short-term memory (LSTM). Meanwhile, the ARMA and ARIMA models were selected for comparative analysis. The findings reveal that the ARMA model exhibits the shortest prediction time, but its RMSE and MAE values are the largest, suggesting that the ARMA model yields the poorest predictive performance. The LSTM model achieved the lowest RMSE, and its MAE closely approached that of ARIMA, but the ARIMA model could only predict the gas pressure in the short term. Therefore, it can be employed as the prediction model for gas pressure in coal and gas outburst experiments. The research findings offer significant auxiliary support for predicting the change of gas pressure during coal and gas outbursts, thereby facilitating further prediction and prevention of such occurrences. [Received: October 9, 2023; Accepted: February 27, 2024]

Keywords: coal and gas outburst; gas pressure prediction; LSTM neural network; multistep prediction.

DOI: 10.1504/IJOGCT.2025.143014

International Journal of Oil, Gas and Coal Technology, 2025 Vol.37 No.1, pp.102 - 124

Received: 06 Oct 2023
Accepted: 27 Feb 2024
Published online: 02 Dec 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article