Underground Mine Flood Accidents Bayesian Network Learning and Inference

Abstract

In order to reduce the flood accidents in underground coal mine of China, a probabilistic inference model with machine learning method is proposed in this paper. Firstly, based on a dataset collected from 99 cases from year 2000 to 2018, the Fault tree analysis (FTA) and causal analysis are applied to determine the related factors of the flood accidents. Then an accurate Bayesian network (BN) model of flood accidents is obtained by structure learning and parameter learning. Finally, the sensitive factors and key factors are acquired by sensitivity analysis and maximum cause chain analysis, according to which the corresponding strategies for reducing flood accidents are put forward to reduce the risk.