基于通勤大数据的出行空间分布模型测评与改进  

Trip Spatial Distribution Model Evaluation and Improvement Based on Commuting Big Data

作　　者：盛志前[1] 吴子啸[1] SHENG Zhiqian;WU Zixiao(China Academy of Urban Planning&Design,Beijing 100044,China)

机构地区：[1]中国城市规划设计研究院,北京100037

出　　处：《城市交通》2023年第1期69-73,107,共6页Urban Transport of China

摘　　要：通勤出行是城市交通早晚高峰的主要构成部分,理解和预测通勤出行空间分布一直是城市交通研究的主要方向之一。基于互联网位置服务所识别的城市通勤大数据,以工程领域广泛应用的重力模型为例,研究和评价出行空间分布模型对通勤出行样本量的敏感性以及在不同空间尺度上对实际通勤出行空间分布的重现能力,并剖析了传统出行空间分布模型所引起的“碎片化”结果及其成因。克服传统方法仅考虑群体统计特征的缺陷,引入反映个体通勤出行延续性的因子,提出了新的通勤空间分布模型及求解算法。新的模型和算法能够更好地适应通勤大数据背景下巨量分析单元的情形,基于实际数据验证了其对通勤出行空间分布的重现能力优于传统出行空间分布模型。Commuting trips are a major component of urban rush hour traffic.Understanding and predicting the spatial distribution of commuting trips has been one of the mainstream in urban transportation research.Based on the urban commuting big data identified by Location Based Service(LBS),taking the widely-used gravity model as an example,this paper evaluates the sensitivity of the trip spatial distribution model to the sample size of commuting trips.The model's capability in reproducing the actual commuting distribution with different spatial scales is also investigated.Based on the analysis of the fragmentation caused by traditional trip spatial distribution models and its underlying causes,a new commuting spatial distribution model and solution algorithm that adapts to massive analysis units in the context of commuting big data are proposed.The predictive capability of the new model and algorithm was verified based on actual data,and the results show that its ability to reproduce commuting distribution is superior to that of traditional trip spatial distribution models.

关 键 词：交通分析与建模 空间分布模型 重力模型 通勤大数据 

分 类 号：U491.12[交通运输工程—交通运输规划与管理]