基于随机森林的高速公路运营隧道风险判别法则研究*

为快速、准确识别山区高速公路隧道不同区段风险因素组合，基于交通事故调研数据，分析其时空变化规律；从人-车-路-环境角度系统筛选风险影响因素，采用随机森林理论构建风险等级预测模型；以决策树与规则相互关系为基础，结合自变量重要评分法，提出基于随机森林的高速公路运营隧道风险判别法则。研究结果表明:判别法则集预测精度较好且运行时间短，能够明确高速公路不同隧道区段事故成因，为预防隧道严重交通事故发生提供参考。

Research on risk discrimination rule of expressway operating tunnel based on random forest

In order to quickly and accurately identify the combination of risk factors in different sections of mountain expressway tunnels,the temporal and spatial variation law was analyzed based on the survey data of traffic accidents.From the perspectives of people,vehicles,roads and environments,the risk influencing factors were systematically screened,and a risk level prediction model was constructed by using the random forest theory.Based on the interrelationship between decision tree and rules,combined with the independent variable important scoring method,the risk discrimination rules of expressway operating tunnel based on random forest was proposed.The results showed that the discriminant rule set had good prediction accuracy and short running time,and clarified the causes of accidents in different tunnel sections of expressway,which provides a research idea for preventing the occurrence of serious traffic accidents in tunnels.