考虑不良天气的山区公路运营耦合风险识别

Risk Identification of Coupled Mountain Road Operations Considering Adverse Weather

关于山区公路运营风险因素识别和控制的问题,现有研究大多基于单一风险因素分析,忽视了不同风险因素之间的关联,较少考虑由于不良天气自身占比而形成的数据不平衡性问题。针对这一问题,选取云南省的某条典型山区公路为研究对象,基于Apriori关联分析算法提出山区公路耦合风险识别模型。首先,设置最小支持度阈值为1%、最小置信度阈值60%及最小提升度阈值1.5,挖掘得到事故总体数据的关联规则;其次,设置最小支持度阈值为2%、最小置信度阈值为60%及最小提升度为2,挖掘得到不良天气条件下事故关联规则;最后,分别选出排名前25%的关联规则,得到针对该山区公路全天候的耦合风险。结果显示,考虑不良天气状况进行山区公路事故风险因素组合关联规则的二次挖掘,避免了总体事故分析结果的失真和潜在的低支持度有效关联规则被过滤掉的问题,二次挖掘得到的高频山区公路事故风险组合形式说明考虑数据不平衡性问题进行二次挖掘的必要性。

Regarding the identification and control of risk factors for mountainous highway operation,most existing studies are based on single risk factor analysis,ignoring the association between different risk factors,and less consideration is given to the problem of data imbalance formed by the small percentage of bad weather itself.To address this problem,a typical mountain highway in Yunnan Province is selected as the research object,and a coupled risk identification model for mountain highways is proposed based on Apriori correlation analysis algorithm.Firstly,the minimum support threshold of 1%,the minimum confidence threshold of 60%and the minimum enhancement threshold of 1.5 are set to obtain the association rules for the overall accident data;Secondly,the minimum support threshold of 2%,the minimum confidence threshold of 60%and the minimum enhancement of 2 are set to obtain the association rules for accidents under adverse weather conditions;Finally,the top 25%association rules are selected to obtain the coupling risk for the mountainous.After that,the top 25%association rules are selected to obtain the coupled risk for the mountainous area allweather road.The results show that the secondary mining of association rules for the combination of risk factors of mountain highway accidents considering adverse weather conditions avoids the distortion of the overall accident analysis results and the problem that the potentially effective association rules with low support are filtered out,and the form of high-frequency mountain highway accident risk combinations obtained from the two mines indicates the necessity of secondary mining considering the problem of data imbalance.