安全与环境工程2025，Vol.32Issue(3)：72-78,100,8.DOI:10.13578/j.cnki.issn.1671-1556.20240190
企业安全生产预警等级划分的阈值确定方法
Method of determining the threshold for early warning levels of enterprise production safety
摘要
Abstract
The Technical Standard of Production Safety Early Warning System for Enterprises in the Metallurgy and Other Industrial and Trade Industries provides a comprehensive interpretation,categorizing the early warning state of enterprise safety production into four levels:safety,attention,warning,and danger.The defined values between each level serve as thresholds for early warnings.By leveraging the enterprise safety management system and conducting big data analysis,a novel approach is established to determine the thresholds for classifying safety production early warnings.Firstly,an early warning index system encompassing all risk factors that may lead to safety production accidents was constructed.Then,the safety production index(SPI)model was formulated using criteria importance through intercriteria correlation(CRITIC).Finally,python programming language aids in employing the naive Bayes classifier algorithm and Gaussian approximation fitting density function for analytical purposes,ultimately determining and grading the corresponding thresholds of the four warning states.Taking Xintian Steel Group as an example demonstrates that four feasible,effective,and practical safety warning levels are calculated:safety(SPI≤0.44),attention(0.44<SPI≤0.55),warning(0.55<SPI≤0.61),danger(SPI>0.61).关键词
安全生产预警指数(SPI)/CRITIC赋权法/贝叶斯分类器/阈值/等级划分Key words
safety production index(SPI)/criteria importance through intercriteria correlation(CRITIC)weighting method/Naive Bayes classifier/threshold/grade classification分类
安全科学引用本文复制引用
丁洪玲,王福生,董宪伟,韩圣君,王达,李国友..企业安全生产预警等级划分的阈值确定方法[J].安全与环境工程,2025,32(3):72-78,100,8.基金项目
中央引导地方科技发展资金项目(236Z5503G) （236Z5503G）
河北省华北理工大学研究生创新能力培养项目(CXZZBS2025156) （CXZZBS2025156）
