摘要
针对利用实时浮动车数据估计路段行程时间时存在的数据稀疏性问题,提出了构建三层神经网络模型,以目标路段与邻接路段间的特征关系为输入、目标路段与邻接路段行程时间比值为输出,利用浮动车历史大数据获取路段之间的交通时空关联关系,继而用于路段行程时间的推断。采用武汉市2014年3~7月的浮动车GPS历史数据进行验证,得到的路段行程时间估计值的平均绝对百分比误差小于25%,证明了所提方法的有效性。
Although thereare massive quantities of floating car GPS data,partial links lack real data during certain periods of time.Therefore,we cannot estimate target link travel time.Consideringthe sparse data problemwhen using floating car data toestimate link travel time,we put forward aninferred method based on big floating car data.We designed a three-layer artificial neural network modelwhose input i and output information are the feature relationship and the travel time ratio between target link and adjacent link,respectively.We obtained spatiotemporal traffic association relationships using big historical floating car data and then inferred link travel times.The model was verified using big historical floating car data for Wuhanfrom March to July,2014.The MAPE of estimated values of link travel time was less than 25% demonstrating the effectiveness of the proposed method.
作者
张发明
朱欣焰
呙维
胡涛
ZHANG Faming ZHU Xinyan GUO Wei HU Tao(State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan 430079, China Collaborative Innovation Center of Geospatial Technology, 129 Luoyu Road, Wuhan 430079, China)
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2017年第1期56-62,共7页
Geomatics and Information Science of Wuhan University
基金
国家863计划(2013AA122301)
国家科技支撑计划(2012BAH35B03)~~
关键词
浮动车大数据
数据稀疏性
时空关联关系特征
神经网络
行程时间推断
big data of floating car
data sparsity
spatiotemporal association relationship features
artificial neural network
travel time inference
