Journal of Arid Meteorology ›› 2020, Vol. 38 ›› Issue (2): 339-345.DOI: 10.11755/j.issn.1006-7639(2020)-02-0339

Previous Articles     Next Articles

Characteristics of Expressway Traffic Accident and Meteorological Warning Model Based on Logistic Regression in Hebei Province

WANG Jie1, QU Xiaoli1,2, ZHANG Jinman1   

  1. 1. Hebei Provincial Meteorological Service Center, Shijiazhuang 050021, China;
    2. Key Laboratory of Weather and Ecological Environment of Hebei Province, Shijiazhuang 050021, China
  • Online:2020-04-28 Published:2020-04-28

河北高速公路交通事故特征及其气象预警模型

王洁1,曲晓黎1,2,张金满1   

  1. 1.河北省气象服务中心,河北 石家庄 050021;
    2.河北省气象与生态环境重点实验室,河北 石家庄 050021
  • 通讯作者: 曲晓黎(1982— ),女,吉林蛟河人,高级工程师,主要从事交通气象应用技术研究. E-mail: hebqx_quxiaoli@126.com。
  • 作者简介:王洁(1988— ),女,工程师,主要从事电力、交通气象服务. E-mail: 1107834478@qq.com。
  • 基金资助:
    河北省重点研发计划项目“高速公路复杂路面高分辨率恶劣天气精准预警技术研究”(19275413D)和河北省气象局创新团队“交通气象服务技术研发及应用”共同资助

Abstract: Based on the traffic accidents data on expressway of Hebei Province and conventional observation data at 109 national weather stations from October 2015 to October 2018, the temporal variation characteristics of expressway traffic accidents and the relationship with meteorological factors in summer and winter half years were contrastively analyzed, firstly. Then, the meteorological factors which had obvious influence on traffic accidents were selected by principal component analysis method and the binary Logistic regression model was introduced to establish the meteorological early warning models of expressway traffic accidents in Hebei Province in summer and winter half years, respectively. Finally, the accuracy of two models was tested. The results show that the diurnal and monthly change characteristics of traffic accidents were obvious on expressway of Hebei Province. The frequency of traffic accident in summer half year was 1.4 times more than that in winter half year, and the monthly fluctuation in summer half year was weaker than that in winter half year. The diurnal variation of traffic accidents in summer and winter half years presented  ‘M’ pattern distribution. The traffic accidents in the daytime were more than that in the nighttime, the peak value appeared at 10:00 BST and 15:00 BST, and the traffic accidents at each time in summer half year (except from 18:00 BST to 20:00 BST) were higher than that in winter half year. The meteorological early warning model of expressway traffic accidents in Hebei integrated humidity, precipitation, wind speed and air pressure factors, while the temperature factor was also introduced in model in winter half year. The prediction accuracy of early warning model was above 99% to the samples less than or equal to the mode of traffic accidents, while that was lower to the samples above the mode, but the overall accuracy was still above 80%, which indicated that the model had a certain reference significance to expressway traffic early warning.

Key words: traffic accidents of expressway, principal component analysis, Logistic regression, model test

摘要: 利用2015年10月至2018年10月河北省高速公路交通事故资料和109个国家站地面常规观测资料,对比分析夏半年和冬半年高速交通事故的时间变化特征及与气象要素的关系,利用主成分分析方法筛选出对交通事故影响较大的气象因子,并引入二元Logistic回归模型,分别建立夏半年和冬半年高速交通事故气象预警模型,并对模型进行精度检验。结果表明:河北高速公路交通事故具有明显的日、月变化特征,夏半年事故量是冬半年的1.4倍,且夏半年事故量月波动较冬半年弱;夏、冬半年交通事故日变化均呈“M”型分布,白天远高于夜间,峰值分别出现在10:00和15:00,且夏半年各时次(18:00—20:00除外)事故量高于冬半年。河北高速交通事故气象预警模型综合了湿度、降水量、风速、气压因子,而冬半年还增加了温度因子,预警模型对低于或等于事故发生众数的样本预测正确率均在99%之上,而对高于事故发生众数的样本错判率较高,但整体精度仍在80%以上,说明模型对高速公路交通预警有一定参考意义。

关键词: 高速公路交通事故, 主成分分析, Logistic回归, 模型检验

CLC Number: