Global sensitivity analysis of roof hazard factors based on information entropy and the surrogate model

Abstract

Abstract The prevention and management of coal mine roof accidents remain challenging issues because it is difficult to evaluate and quantify the interaction effects of the disaster hazard factors objectively. This paper proposes a novel approach: combining information entropy and the surrogate model—and applies Sobol’s method, aiming to solve it and to obtain the hazard factors’ 1th and the global sensitivity value without human intervention. The results show that: (1) The complex logical relationships and interactions of roof hazard factors can be transformed into quantifiable numerical values by building a co-occurrence matrix of disaster factors and calculating its information entropy. (2) The sensitivity levels of roof hazard factors can be successfully distinguished and categorized into priority management and prevention or general management and prevention using the surrogate model and Sobol’s sensitivity method. The novel sensitivity analysis approach suggested in this study considers both the individual impacts of hazard factors and their interactions, offering a more thorough framework for risk assessment as well as a fresh perspective and tool for coal mine safety research.