基于序列比对的行人过街风险识别研究

行人过街序列反映了行人-机动车的行为交互在时空的连续性，对行人过街序列进行风险识别并对其关联影响因素进行分析有助于评估不同情景下行人过街行为的风险性. 本文对行人-机动车在交叉口区域内的碰撞事故数据进行挖掘分析.借鉴生物信息学的序列比对方法，对行人过街事故序列进行相似性比对及聚类分析，并基于灰色关联方法分析事故序列的风险耦合.结果表明，行人-机动车在交叉口的风险事故序列主要分为6种，Seq2与Seq5，Seq3与Seq4的事故诱因存在相似性；天气不是Seq1的事故诱因，Seq2与Seq5的事故主要归因于驾驶员.研究结果对完善行人交通风险管理有一定的理论价值和实践意义.

Identify Risks of Pedestrian Crossing Based on Sequence Alignment Method

The sequence of pedestrian crossing reflects the continuity of pedestrian-motor vehicle behaviorals interaction in space-time. Identifying the risks of pedestrian crossing sequences and analyzing the related influencing factors are helpful to evaluate the risks of pedestrian crossing behaviors under different circumstance. The collision accidents data of pedestrian- motor vehicles in the intersection area are excavated. By using the sequence alignment method of bioinformatics, the similarity of pedestrian crossing accidents sequences is compared and classified. Furthermore, the risks coupling influences factors of the sequences are analyzed based on gray relational method. The results show that, six kinds of risk sequences of pedestrian crossing accidents are identified. The causes of collisions are similar between Seq2 and Seq5 or Seq3 and Seq4. Otherwise, weather is not the main cause of Seq1, and the cause of collisions of Seq2 and Seq5 lies mainly in the drivers. The research results have certain theoretical value and practical significance to improve pedestrian traffic risk management.