摘要
共享单车是一种便宜、绿色环保的短途出行工具,已经成为缓解城市交通压力的重要方式.对于无桩共享单车,用户无需将自行车归还至停车桩,但这种类型的共享单车在高峰时间可能会过于拥挤.本文提出了一种共享单车停车拥挤区域识别的方法.具体来说,以某市某品牌共享单车为例,首先对共享单车数据进行预处理,然后使用GeoHash算法处理经纬度坐标信息并计算判断共享单车开关锁订单属于哪个停车围栏,采用HDBSCAN(hierarchical density-based spatial clustering of application with noise)聚类算法将停车围栏聚类为停车区域,在此基础上提出了基于“留存流量与留存密度的综合指标”的方法识别停车拥挤区域.通过分析,识别出的停车拥挤区域符合实际情况.所提出的停车拥挤区域识别方法能够为“削峰填谷”引导调度提供有效的数据支持,给共享单车企业提供一定的参考.
Shared bikes have become the key to alleviating urban traffic pressure and can be inexpensively and environmentally friendlily used for short-trip transportation within metropolis.The dockless bikes are designed for situations in which users do not need to return the bike to a kiosk.However,they tend to clog up busy squares and thoroughfares.In this paper,a method for identifying crowded parking areas of shared bikes is proposed.Specifically,we take a certain brand of shared bikes in a typical city as an example.Shared-bike data are pre-processed first,and then the GeoHash algorithm is used to process the latitude and longitude coordinate information.Afterwards,we calculate to determine which parking fence the shared-bike switch lock order belongs to.The HDBSCAN clustering algorithm is used to cluster parking fences as parking areas.On this basis,a method based on"the comprehensive index of retained traffic and retained traffic density"is proposed to identify those crowded parking areas.Experimental results show that these identified crowded parking areas coincide with the actual situation.The proposed method for identifying crowded parking areas can provide effective data support for"peak shaving and valley filling"guidance,and serve as a reference for bike-sharing companies.
作者
洪文兴
陈明韬
刘伊灵
朱嘉诚
王明磊
HONG Wenxing;CHEN Mingtao;LIU Yiling;ZHU Jiacheng;WANG Minglei(School of Aerospace Engineering,Xiamen University,Xiamen 361102,China;School of Mathematical Sciences,Xiamen University,Xiamen 361005,China;School of Software,Beihang University,Beijing 100083,China)
出处
《厦门大学学报（自然科学版）》
CAS
CSCD
北大核心
2022年第6期1030-1037,共8页
Journal of Xiamen University：Natural Science
