为细化分析高速公路车辆出行特征,提升高速公路管控效率及精细化服务水平,提出“先定义后聚类”的高速公路出行车辆群体分类模型。基于重庆市某高速公路特定通道收费数据,以小客车为研究对象,首先,在分析车辆整体特征的基础上,提取并定义研究时段内单次出行或仅在周末出行的车辆;然后,以自然周为周期,从出行强度、出行时间维度、出行空间维度3方面构建剩余车辆出行量化指标;最后,利用K-means++算法对车辆聚类,综合地域实际背景将出行车辆分为6类,其中70%为“单次出行”或“零星出行”车辆,“通勤”及“营运”车辆只占车辆总数的0.82%,且“通勤”车辆多在20km以内的OD上出行。研究表明,所提出的模型能将高速公路出行车辆有效分类,且各类群体的出行时空分布存在较明显差异,可为针对性的交通需求管理提供合理依据。
In order to analyze the travel characteristics of vehicles travelling on expressways and improve the efficiency of expressway management and level of refined service, a group classification model of expressway vehicles based on " define before clustering" was proposed. Based on the toll data of a certain expressway in Chongqing, the passenger cars were taken as the research objects. Firstly, the overall characteristics of those vehicles were analyzed and the single-trip or weekend-only vehicles in the study period were extracted and defined. Then, taking the natural week as the cycle, the travel quantitative indexes for the remaining vehicles were constructed from aspects of travel intensity, travel time, travel space. Finally, K-means++ algorithm was applied to cluster the vehicles. They were divided into 6 categories combined with the actual regional background. 70% of the vehicles were single-trip or sporadic-travel vehicles, while commuter and commercial vehicles only accounted for 0.82% of the total number of vehicles, and most of the commuter vehicles travelled on OD within 20km. The results show that the proposed model can effectively classify the vehicles travelling on expressway, and there are obvious differences in the travel time and space distribution of various vehicle groups, which can provide a reasonable basis for targeted traffic demand management.
[1] 岑晏青,宋向辉,王东柱,等. 智慧高速公路技术体系构建[J]. 公路交通科技,2020,37(7):111-121.
[2] 袁佳,杨东援,孙世超,等. 基于公交IC卡数据与问卷调查数据的公交改善决策支持[C]// 2017年中国城市交通规划年会论文集. 北京:中国建筑工业出版社,2017:1403-1414.
[3] 刘家玮. 北京地铁分时定价对乘客出行行为影响研究[D]. 北京:北京交通大学,2018.
[4] 席洋. 基于AFC数据的城市轨道交通乘客聚类及客流分配研究[D]. 北京:北京交通大学,2019.
[5] Kieu L M, Bhaskar A, Chung E. Passenger Segmentation Using Smart Card Data[J]. IEEE Transactions on Intelligent Transportation Systems, 2015, 16(3): 1537-1548.
[6] Briand A S, C?me E, Mahrsi M, et al. A Mixture Model Clustering Approach for Temporal Passenger Pattern Characterization in Public Transport[J]. International Journal of Data Science & Analytics, 2016(1): 37-50.
[7] 吕红霞,王文宪,蒲松,等. 基于聚类分析的铁路出行旅客类别划分[J]. 交通运输系统工程与信息,2016,16(1):129-134.
[8] 畅玉皎,杨东援. 基于车牌照数据的通勤特征车辆识别研究[J]. 交通运输系统工程与信息,2016,16(2):77-82,112.
[9] 邢雪,于德新,田秀娟,等. 基于数据挖掘的高速公路行程时间预测[J]. 华中科技大学学报(自然科学版),2016,44(8):36-40.
[10] 陈家炎. 基于联网收费数据的高速公路交通流特征分析[D]. 广州:华南理工大学,2014.
[11] 廖浩霖,翁小雄. 基于高速收费数据的路径速度还原与流量推断[J]. 公路工程,2020,45(1):50-54,113.
[12] 钱超,杨孟,耿健,等. 高速公路ETC客户细分方法研究[J]. 交通运输系统工程与信息,2018,18(4):231-239.
[13] 魏广奇,苏跃江,吴德馨,等. 基于高速公路流水数据的通勤车辆特征研究[J]. 交通运输系统工程与信息,2019,19(3):237-244.
[14] Arthur D, Vassilvitskii S. K-means++: The Advantages of Careful Seeding[C]// Proceedings of the Eighteenth Annual ACM-SIAM Symposium on Discrete Algorithms. New Orleans: Society for Industrial and Applied Mathematics, 2007: 1027-1035.
[15] 刘旭东,邹纪元. 基于联网收费数据的高速公路出行特征分析[J]. 综合运输,2019,41(10):113-117.
[16] 唐艳. 基于高速公路收费数据的疑似非法客运车辆辨识方法研究[D]. 重庆:重庆交通大学,2019.