首页|危险货物运输驾驶人风险倾向分类及识别模型研究

危险货物运输驾驶人风险倾向分类及识别模型研究

Research on risk tendency classification and identification model of drivers in dangerous goods transport

扫码查看
原文链接
  • NETL
  • NSTL
  • 维普
  • 万方数据
为合理评估危险货物运输驾驶人驾驶过程中的风险倾向,建立危险货物运输驾驶人风险倾向聚类及辨识体系,以动态监控系统中记录的驾驶人实时违规预警数据为基础,选取可能引发交通冲突的安全关键事件为特征参数,利用探索性因子分析方法实现指标降维,提取驾驶人风险倾向主因子,并通过K-means算法聚类不同风险倾向的驾驶人,最后基于聚类结果监督训练随机森林模型,辨识未知驾驶人的风险倾向.结果表明,利用选取的8类安全关键事件特征参数,可以将驾驶人风险倾向划分为攻击驾驶倾向、鲁莽驾驶倾向、驾驶分神倾向和驾驶疲劳倾向,且可以识别风险较低的驾驶人,基于随机森林模型的驾驶人风险倾向识别准确率为88.68％,可以较好地实现危险货物运输驾驶人风险倾向辨识.研究结果为危险货物运输驾驶人风险倾向分类及识别提供了方法依据.
Transporting hazardous materials is a dynamic and dangerous procedure.A risk tendency accumulation and identification model for drivers of risky commodities is designed to allow for a reasonable evaluation of the risk tendency of drivers during the transportation process.Based on the real-time violation warning data of drivers recorded in the dynamic monitoring system and the Safety-Critical Events(SCEs)that might cause traffic conflicts are chosen as characteristic parameters.The indicators'dimension is reduced using the empirical factor analysis method,which also helps to identify the major causes of drivers'propensity for risk.The K-means algorithm is employed to group drivers with various risk propensities.Based on the clustering findings,a random forest model is lastly trained to determine the risk tendency of unknown factors.By contrasting the classification outcomes of several models on the unbalanced dataset,the classification performance of the random forest model is assessed.The findings indicate that drivers'risk tendencies can be classified into four categories based on the violation warning data gathered by the dynamic monitoring system and the chosen eight safety-critical event characteristic parameters:contentious driving tendency,reckless driving tendency,driving distraction tendency,and driving fatigue tendency.Drivers with varying risk tendencies and drivers with higher levels of safety can be efficiently classified using the K-means algorithm.The silhouette coefficient and the sum of squared errors indices are used to assess the clustering effect.The ideal cluster count is calculated as 5.Based on the random forest model,88.68％of driver risk inclinations can be accurately identified.This result shows that the random forest model can accurately identify the risk tendency of unidentified drivers in risky goods vehicles.It is important to note that the random forest model outperforms the Support Vector Machine(SVM)and BP neural network models in classification for unbalanced datasets.The research results provide a method basis for classifying and identifying drivers'risk tendencies in dangerous goods transport.They also provide feasible suggestions for further improving driver safety levels.

safety engineeringdangerous goods transportationdriverrisk tendencySafety-Critical Events(SCEs)exploratory factor analysisRandom Forest(RF)

沈小燕、韩小强、羊家豪、郭丹、陈煜、董相勇

展开 >

长安大学汽车学院,西安 710064

北京工业大学理学部,北京 100124

陕西省道路运输事业发展中心,西安 710003

南京交投信息技术有限公司,南京 210009

展开 >

安全工程 危险货物运输 驾驶人 风险倾向 安全关键事件(SCEs) 探索性因子分析 随机森林(RF)

国家重点研发计划项目陕西省交通运输厅交通科研项目

2019YFE010800020-19R

2024

10.13637/j.issn.1009-6094.2023.0706

安全与环境学报
北京理工大学 中国环境科学学会 中国职业安全健康协会

安全与环境学报

CSTPCD北大核心
影响因子：0.943
ISSN：1009-6094
年,卷(期)：2024.24(4)
  • 24