基于手机信令数据的交通方式识别模型  

Traffic mode recognition model based on mobile phone signaling data

作　　者：宋丽华[1] 于磊 张扬[1] 马东超[1] SONG Li-hua;YU Lei;ZHANG Yang;MA Dong-chao(School of Information Science and Technology,North China University of Technology,Beijing 100144,China)

机构地区：[1]北方工业大学信息学院,北京100144

出　　处：《计算机工程与设计》2024年第11期3441-3447,共7页Computer Engineering and Design

基　　金：北京市教育委员会科学研究计划基金项目(KM202110009002);北京自然-教委联合重点基金项目(KZ201810009011)。

摘　　要：为解决现有研究在区分居民交通方式方面存在的问题,提出一个基于手机信令数据的居民交通方式分类模型。模型数据清洗部分采用滑动窗口方式提取3类特征训练模型,成功检测出80%以上的噪声数据;交通识别部分划分为3个二分类问题,即机动车与非机动车、步行和骑行、公交车和私家车。模型利用自主采集的信令数据,结合路网、导航数据和站点匹配算法,设计出居民交通方式分类模型。实验结果表明,与其它方法相比,公交车和私家车的分类准确率达到94%,4种分类结果的整体准确率达到了86.76%,验证了该设计的准确性和可行性。To address the existing issues in distinguishing residents’transportation modes,a resident transportation mode classification model based on mobile signaling data was proposed.A sliding window approach was used in the data cleaning section to extract three categories of features to train the model,successfully detecting more than 80%of the noise data.The transportation mode identification part was divided into three binary classification problems including motor vehicles vs.non-motor vehicles,walking and cycling,and buses vs.private cars.The autonomously collected signaling data was used,while the road network,navigation data,and station matching algorithms were combined,a resident travel mode classification model was designed.Experimental results show that,compared with other methods,the classification accuracy of buses and private cars reaches 94%,and the overall accuracy of the four classification results reaches 86.76%,verifying the precision and feasibility of the proposed design.

关 键 词：城市交通 交通方式识别 数据清洗 信令数据 站点匹配 出行链特征提取 轨迹数据 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]