Regional Prediction of Coal and Gas Outburst Under Uncertain Conditions Based on the Spatial Distribution of Risk Index.

The hazard of coal and gas outburst underground is extremely high, with complex influencing factors and uncertain risk categorization. In this case, the traditional regional outburst prediction methods fail to fulfill the demands of spatial risk quantitative distribution, complicating the implementation of targeted regional outburst prevention measures. By combining the matter–element cloud model (MECM) with improved Dempester–Shafer evidence theory, this work proposes a novel regional outburst prediction approach based on the spatial distribution of risk index. First, a 14-indicator risk system (denoted as GCCG in this paper) is constructed using the 50 percent coverage cross-unit division approach from four categories, including gas occurrence, characteristics of coal body, coal seam occurrence structure, and ground stress and concentration, and each indicator, as an independent evidence source, generates basic probability assignments by MECM, which takes into account the contribution of the internal attribute of evidence sources to fusion results through subjective and objective weights. Then, high conflict evidence is filtered using the coefficient K. The Jousselme distance and Euclidean distance (ED) were employed to increase the reliability of evidence sources, thus obtaining comprehensive risk fusion results. The improved Centre of Distribution criterion, which is more suitable for outburst prediction, is developed based on the Centre of Distribution criterion to achieve scientific quantification of outburst risk distribution. Finally, the validity and feasibility of such a prediction method are confirmed by collecting parameters of 294 units divided by field examples, concluding that the improved evidence source of the ED method better satisfies the demand for outburst prediction. [ABSTRACT FROM AUTHOR]