考虑时空自相关性的共享电动汽车出行选择影响因素分析  

Analysis of factors influencing shared electric vehicle travel choice considering spatiotemporal autocorrelation

作　　者：廖洋 罗霞[1,2] 王红杰 LIAO Yang;LUO Xia;WANG Hongjie(School of Transportation and Logistics,Southwest Jiaotong University,Chengdu 611756,China;National Engineering Laboratory of Integrated Transportation Big Data Application Technology,Chengdu 611756,China)

机构地区：[1]西南交通大学,交通运输与物流学院,成都611756 [2]综合交通大数据应用技术国家工程实验室,成都611756

出　　处：《交通运输工程与信息学报》2024年第4期67-84,共18页Journal of Transportation Engineering and Information

基　　金：四川省科技计划项目(2024NSFSC0179)。

摘　　要：共享电动汽车作为新兴的低碳出行方式,探究其时空分布规律及影响因素,有助于优化交通资源配置,推动城市交通系统的可持续发展。本文首先研究了共享电动汽车订单量的时空自相关特征;其次,基于对传统区域聚合方法的改进,结合对兴趣点(POI)空间分布模式(密度、邻近性、集聚性)的精细化定义,本文建立了综合考虑时空特征、站点属性及建成环境特征的“S-T+S+T+5D”影响因素量化体系;最后,通过构建并对比多个识别模型,本文揭示了各因素的影响程度和作用机理。结果表明:(1)相较于广义线性模型及随机森林模型,考虑地理交互效应的广义可加混合模型能更准确地识别影响因素,有效解释了服务需求的空间依赖特征。(2)影响因素中,站点属性的影响程度最高;车位容量表现出阈值效应,即区域容量高于70会抑制出行需求;站点间的过近距离会加剧内部竞争。(3)共享电动汽车在补充城市公共交通服务不足方面具有一定潜力,尤其在地铁入口2 km外的服务薄弱区域和大型客运枢纽周边。(4)土地混合利用强度呈现倒U型的非线性关系,娱乐场所密度、高校邻近性和医疗场所集聚性均对共享电动汽车出行产生促进作用。研究结果可为运营商进行短期站点优化及长期布局规划提供理论支撑。As an emerging low-carbon mode of transportation,shared electric vehicles(SEVs)have the potential to optimize traffic resource allocation and promote the sustainable development of urban transportation systems.Therefore,it is critical to study their spatiotemporal distribution patterns and influencing factors.This study investigated the spatiotemporal autocorrelation characteristics of SEV trips.It then integrated an improved regional aggregation method with a refined definition of Points of Interest(POIs)spatial distribution patterns,which include density,proximity,and agglomeration.A comprehensive“S-T+S+T+5D”quantification system for influencing factors was established,which considered spatiotemporal features,station attributes,and built environment characteristics.Finally,by constructing and comparing multiple identification models,this study revealed the extent and mechanisms of various influencing factors.The results indicate the following:(1)Compared to the Generalized Linear Model(GLM)and Random Forest(RF)models,the Generalized Additive Mixed Model(GAMM),which considers geographical interaction effects,can more accurately identify influencing factors and effectively explain the spatial dependency of service demand;(2)Among the influencing factors,station attributes have the most significant impact;parking capacity exhibits a threshold effect,where an area capacity exceeding 70 units can suppress travel demand,and overly close proximity between stations can intensify internal competition;(3)SEVs have the potential to compensate for insufficient urban public transportation services,particularly in areas with weaker service coverage more than two kilometers from metro entrances and around major passenger transport hubs;(4)The intensity of land-use mix exhibits an inverted U-shaped nonlinear relationship,and the density of entertainment venues,proximity to universities,and agglomeration of medical facilities all promote SEV travel.These findings provide theoretical support for operators when conducting sho

关 键 词：城市交通 非线性分析 广义可加混合模型 共享电动汽车 时空自相关 

分 类 号：U121[交通运输工程]