基于贝叶斯网络的煤矿事故人因推理

Human factors inference of safety accidents in coal mine based on Bayesian network

煤矿事故的不可重现性决定了事故原因的调查具有很强的不确定性,如何通过事故发生后的相关信息提高事故深层次原因调查的准确性是非常重要的。将HFACS与贝叶斯网络相结合,以煤矿事故HFACS分析结果为样本,通过卡方检验和让步比分析建立人因的贝叶斯网络因果图,进一步利用最大似然估计算法确定了煤矿事故人因的贝叶斯网络参数。最后,以双柳煤业顶板事故的调查信息为证据推理导致煤矿事故发生的深层次原因,提高事故原因调查的准确性,从而验证模型的有效性。

The irreproducibility of accidents in coal mine leads to the uncertainty of investigations on accidents causes,which makes it very important to find out the deep reasons of accidents accurately by using the relevant information collected after the accidents.Combining the human factors analysis and classification system（HFACS） with Bayesian Network（BN）,taking HFACS analysis results of coal mine accidents happened before as samples,the Chi-square test and odds ratio analysis were used to establish the Bayesian network causality diagram of human factors,and the maximum likelihood estimation algorithm was used to determine the Bayesian network parameters of human factors.Finally,based on survey information of the roof accident in Shuangliu Coal Industry Group,the deep reasons of accident were speculated,and the validity of the model was verified because of the accuracy improvement of accident investigation.