摘要
应用上海市高速公路1 104条事件数据,基于专家知识和数据融合方法建立贝叶斯网络结构;利用服从Dirichlet分布的贝叶斯方法进行参数学习;运用团树传播算法进行推理分析.研究了上海市高速公路尾随相撞事件类型与不同道路环境条件之间的关系.在验证贝叶斯网络模型的有效性后,系统分析事件致因,并提出改进措施.发现重大尾随相撞事件易发生在大中型车与小型车之间;夜间易发生大中型货车的重大尾随相撞事件,尤其是凌晨0时至6时;路表潮湿状态下的非普通路段上易发生大中型客车的重大尾随相撞事件.结果表明贝叶斯网络建模能够更好的反映事件致因因素的多维性及关联性,是交通事件致因分析的有效方法.
By the use of 1104 recorded Shanghai highway incidents ,the topological structure of BN （Bayesian network） was formed with references to expert knowledge and data fusion method .Bayes-ian method was used to complete the process of parameter learning with Dirichlet prior distribution . Network analyzes were inferred using Clique tree propagation algorithm engine .The paper studied the relationship between rear-end collision and different road environment .After Verifying the effective-ness of the Bayesian network model ,the causation of incident was analyzed and improvement measures were proposed .Inference results indicate that significant rear-end collision easily occurs between large and medium-sized trucks and small car .large and medium-sized trucks easily happen to significant rear-end collision in the evening ,especially from 0 ：00 to 6：00 .Significant rear-end collision easily oc-curs to the wet road surface of unordinary road by large and medium-sized coaches .Bayesian network is the useful method of traffic incident cause analysis to reflect the multidimensional and associated na-ture of the incidents causation .
出处
《武汉理工大学学报（交通科学与工程版）》
2014年第1期111-115,共5页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
国家自然基金项目资助(批准号:51078270)
关键词
交通安全
尾随相撞
贝叶斯网络
致因分析
K2算法
团树传播算法
traffic safety
rear-end collision
Bayesian network
causation analysis
K2 algorithm
clique tree propagation algorithm
