doi: 10.12013/qxyjzyj2019-021 |
沪昆高速江西段交通事故气象风险分析和概率预测 |
Probability prediction of meteorological risk for traffic accidents in Jiangxi section along Shanghai-Kunming Expressway |
摘要点击 1334 全文点击 440 投稿时间:2019-04-11 修订日期:2019-05-27 |
查看HTML全文 查看全文 查看/发表评论 下载PDF阅读器 |
基金: 2017年江西省气象局重点项目(编号:17302005151714) |
中文关键词: 交通事故,logistic回归模型,概率预测,沪昆高速 |
英文关键词: traffic accidents logistic regression model probabilistic forecast Shanghai-Kunming Expressway |
|
引用:周雨,杨华,刘志萍,邓德文.2019,沪昆高速江西段交通事故气象风险分析和概率预测[J].气象与减灾研究,42(2):133-139 |
<
中文摘要: |
利用2015—2017年沪昆高速江西段的交通事故资料和沿线气象观测数据,分析了沪昆高速江西段发生交通事故的天气类型和时空分布特征,建立高敏感天气条件下山区型和平原型高速公路交通事故气象风险概率预测模型。结果表明:2015—2017年沪昆高速万辆车流交通事故日变化表现为显著的单峰型,峰值出现在05时;事故高发月份主要集中在4—6月;事故发生地域性强。除日平均气温外,交通事故与同期的气象条件密切相关。采用逻辑回归方法,分别建立了山区型和平原型高速公路交通事故气象风险概率预测模型,该模型预测准确率达到780%和799%。进一步运用该模型对2018年1—6月沪昆高速发生的交通事故加以验证,对山区和平原路段高速交通事故预测的准确率达到7056%和8637%,预测效果较为理想。 |
Abstract: |
Based on the traffic accident data of Shanghai-Kunming Expressway in Jiangxi section from 2015 to 2017 and the meteorological observation data,the weather types and spatio temporal distribution characteristics of highway traffic accidents were discussed, and a meteorological risk probability prediction model for highway traffic accidents of mountainous and flatland regions during the severe weather conditions were established. The results showed that: diurnal variation of traffic accidents of 10 000 vehicles in Shanghai-Kunming Expressway from 2015 to 2017 presented a significant single peak at 05:00 BT, and high occurrence of traffic accidents were concentrated during April and June. The traffic accidents were closely related to meteorological conditions during the same period, except the average daily temperature. Using logical regression method, the meteorological risk probability prediction models for mountainous type and flatland type highway traffic accidents were established, respectively, the accuracy of which can reach 78% and 799%. The traffic accidents occurred in Shanghai-Kunming Expressway from January to June in 2018 were also applied to test and verify the model, and the accuracy for mountainous type and flatland type highway traffic accidents reached 7056% and 8637%. |