Prediction of coal mine risk based on BN-ELM: Gas risk early warning including human factors

Abstract

Addressing the challenge of integrating quantitative risk data with qualitative behavioral risk information in coal mine safety production, this study, taking gas risk as an example, proposes a BN-ELM (Bayesian Network-Extreme Learning Machine) prediction and early warning method that incorporates behavioral information. By uniformly quantifying behavioral risks and gas data, optimizing model parameters, and integrating control chart technology, this method constructs a coal mine safety situation awareness model. Experimental results demonstrate that this approach significantly reduces prediction errors in gas data (by 0.007), risk values (by 0.01), and safety situation values (by 0.03). This study innovatively considers behavioral risk factors, providing coal mine enterprises with efficient risk management methods and practical tools.