Tu Qiang, Zong Yijiang, Li Zequan, Yue Liang
School of Transportation Engineering, Jiangsu Vocational Institute of Architectural Technology, Xuzhou, China.
School of Management, North China Institute of Science and Technology, Beijing, China.
Sci Rep. 2025 Sep 1;15(1):32183. doi: 10.1038/s41598-025-16721-y.
Most coal mining enterprises in China have established and use safety production information systems for hazard identification and management, but related accident hazard data have not been fully utilized. This study is based on the classification standards defined by the "coal mine major accident hazard determination standards" implemented by the Ministry of Emergency Management in 2021. We constructed a classification system including 15 major hazard categories and 79 minor hazard categories, which served as sample labels for major coal mine accident hazards. The hybrid convolutional neural network (CNN)-transformer model was used to perform hierarchical text classification on the coal mine major accident hazard data, with the bidirectional encoder representations from transformers (BERT) model used as a baseline for comparison. The results show that in the major hazard category classification experiments, the hybrid CNN-transformer model outperformed the BERT model by 3% points in terms of accuracy, recall, and F1 score. In the minor hazard category classification experiments, the hybrid CNN-transformer model achieved a maximum classification performance of 98%, generally exceeding the BERT model. The coal mine accident hazard classification algorithm based on the hybrid CNN-transformer model demonstrates significant classification effectiveness, providing efficient and rapid input support for coal mine major accident hazard identification systems. Compared to existing BERT models, the hybrid CNN-transformer model significantly improves classification accuracy and training efficiency by combining the extraction of local and global features, exhibiting higher stability and classification performance. Compared with the baseline BERT model, the proposed hybrid CNN-transformer achieves up to 3-4% higher F1 while converging faster and using fewer computational resources. Deployed in a cooperating minesafety platform, it has already reduced manual hazard-triage workload by ~ 30% in daytoday operations.
中国大多数煤矿企业已建立并使用安全生产信息系统进行危害识别与管理，但相关事故危害数据尚未得到充分利用。本研究基于应急管理部2021年实施的《煤矿重大事故隐患判定标准》所定义的分类标准。我们构建了一个包含15个主要危害类别和79个次要危害类别的分类系统，作为煤矿重大事故隐患的样本标签。采用混合卷积神经网络（CNN）-Transformer模型对煤矿重大事故隐患数据进行层次文本分类，并将基于变换器的双向编码器表征（BERT）模型作为比较基线。结果表明，在主要危害类别分类实验中，混合CNN-Transformer模型在准确率、召回率和F1分数方面比BERT模型高出3个百分点。在次要危害类别分类实验中，混合CNN-Transformer模型实现了98%的最高分类性能，总体上超过了BERT模型。基于混合CNN-Transformer模型的煤矿事故隐患分类算法显示出显著的分类效果，为煤矿重大事故隐患识别系统提供了高效快速的输入支持。与现有的BERT模型相比，混合CNN-Transformer模型通过结合局部和全局特征的提取，显著提高了分类准确率和训练效率，表现出更高的稳定性和分类性能。与基线BERT模型相比，所提出的混合CNN-Transformer在收敛速度更快且使用更少计算资源的情况下实现了高达3%-4%的更高F1分数。部署在合作的矿山安全平台上，它已经在日常运营中将人工隐患分类工作量减少了约30%。
