Research on Risk Assessment of Coal and Gas Outburst During Continuous Excavation Cycle of Coal Mine with Dynamic Probabilistic Inference

Guorui Zhang,Enyuan Wang,Xiaofei Liu,Zhonghui Li

DOI: https://doi.org/10.1016/j.psep.2024.08.054

IF: 7.8

2024-01-01

Process Safety and Environmental Protection

Abstract:Coal and gas outbursts are a major dynamic hazard in underground coal mining, and predicting these events challenging due to their complex mechanisms. This study introduces an advanced multistate dynamic probabilistic system designed to evaluate the risk of outbursts prior to each excavation cycle. Through the development of a Bayesian network (BN), we establish a dynamic model that captures the complex interactions among essential variables. The novel integration of time slices facilitates the continuous assessment across successive mining cycles. This study tackles the challenge of maintaining network parameter reliability in the static prediction conditions of outburst-prone coal mines. Specifically, in order to overcome the asynchronism of variables testing process, the expert knowledge integration of expectation-maximization parameter learning and fuzzy set theory (FST) is built inside the system. Finally, we present a case study illustrating the application this system across fifty-one consecutive cycles, with regular updates to variable states, showcasing the enhanced predictive capability of Dynamic Bayesian Networks (DBN) over traditional BNs for ongoing risk evaluation mining operations. Additionally, by integrating backward diagnosis with sensitivity analysis, our approach simplifies the identification of key risk factors, offering valuable insights for improving outburst prevention measures and ensuring the safety of coal mining activities.