Quantitative evaluation of the indexes contribution to coal and gas outburst prediction based on machine learning

Abstract

The coal and gas outburst is a most destructive coal mine power disaster, which seriously threatens the safety of coal mine production. Accurate prediction of gas outbursts can effectively reduce its harm, but the mechanism of outburst is still unclear. We propose a gas outburst prediction method based on comprehensive index and machine learning. XGBoost parameter optimization is studied based on grid search and meta-heuristic algorithm optimization to obtain the best parameter combination for outburst prediction. The prediction accuracy of XGBoost with parameters optimization for the outbursts in the test set was 100%. Finally, we have studied the contribution rate of each index in outburst prediction based on XGBoost interpretability, and the gas content has the highest prediction decision-making contribution, which is 28.61%. This study is of great significance to the safety production of coal mines, and enriches the theory of coal and gas outburst.