基于手机信令的城市机动化方式细分双层模型研究

Two-layer Model to Distinguish Urban Motorized Travel Mode Based on Mobile Phone Signaling Data

针对现阶段手机信令数据难以适用城市复杂出行环境,无法有效区分密集路网下机动化出行方式,本文提出一种考虑路径精准拟合与多维时空特征的双层识别模型。在出行路径识别层面,Savitzky-Golay(S-G)滤波能有效平滑信令数据相对实际出行路径的波动,线性插值算法能弥补信令数据时空缺失。在出行方式识别层面,探究了识别路径相似度、出行时间相似度、加速度、小波速度等关键因素,利用K-临近算法识别公交、小汽车。结果表明:本文提出方法能有效细分城市密集路网环境下的公交与小汽车出行,识别准确度分别达到88.29%和82.28%。在不同出行距离、出行时段、拥挤状态、道路等级、道路类型及识别路径相似度等角度,识别效果均优于随机森林等算法。研究支撑了基于信令数据的出行特征精准挖掘,为道路规划建设,公交线网规划等提供重要基础。

It is always difficult to apply the mobile phone signaling data in the actual urban complex travel environment and also challenging to distinguish the motorized travel mode under dense road networks.This paper proposes a twolayer model considering accurate path fitting and multidimensional spatio-temporal characteristics.At the level of travel path identification,S-G filtering can effectively smooth signaling data fluctuation relative to the actual travel path.The linear interpolation algorithm can fill in the time and space gaps.At the level of travel mode recognition,the key factors are mined,including the similarity of the recognized travel path,travel time similarity,acceleration,and wavelet velocity.The K-nearest neighbor algorithm is used to identify the travel modes(by car or bus).The results show that the proposed method can effectively identify the bus and car in the dense urban road network,and the accuracy rates can reach 88.29%and 82.28%,respectively.Under different travel distances,travel time,congestion conditions,road classes,road types,and path similarity,the proposed method is better than the existing algorithms,such as random forest,in the accuracy rate.The research supports the accurate recognition of travel characteristics based on mobile phone signaling data.It also provides an essential basis for road planning,construction,and public transit network planning.