基于BERT的煤矿事故风险LEC评价与优化研究

Research on LEC Evaluation and Optimization of Coal Mine Accident Risk Based on BERT

为了提高煤矿事故安全评价的准确性和客观性,针对煤矿安全风险评估中常用的LEC评价法可能出现的不客观以及评价过程中L值评价过于复杂的现状,提出了一种基于BERT预训练模型代替人工进行打分并对L值的评价指标进行优化的方法。通过专家打分的事故分析对模型进行训练,得到预测模型,利用预测模型对需要评价的数据进行预测打分,再结合优化后的L值公式得到事故发生可能性的分值。研究结果表明:BERT模型在经过训练后预测效果较好,与专家判断指标的综合重合度高达92.09%,且改良后的L值判断公式综合了环境物体因素、施工操作人员因素和安全管理因素,在判断时可以较为客观地体现出潜在风险发生的可能性,该研究为煤矿行业中作业条件的危险性评估提供了一种新的思路和方法。

In order to improve the accuracy and objectivity of coal mine accident safety evaluation,aiming at the current situation that the LEC evaluation method commonly used in coal mine safety risk as⁃sessment may be not objective and the L value evaluation in the evaluation process is too complicated,a method based on BERT pre-training model instead of manual scoring and optimizing the evaluation index of L value is proposed.The prediction model is obtained by training the model through the accident analysis of expert scoring.The prediction model is used to predict and score the data to be evaluated,and then the score of the possibility of accident occurrence is obtained by combining the optimized L value formula.The results show that the BERT model has a good prediction effect after training,and the comprehensive coinci⁃dence degree with the expert judgment index is as high as 92.09%.The improved L-value judgment formula integrates environmental object factors,construction operator factors and safety management factors,and can objectively reflect the possibility of potential risks in the judgment.This study provides a new idea and method for the risk assessment of operating conditions in the coal mine industry.