| 考虑空间相关性的行人宏观安全研究 |
| |
| 引用本文： | 王雪松,杨俊广,杨筱菡,周清雅. 考虑空间相关性的行人宏观安全研究[J]. 中国公路学报, 2018, 31(5): 136-142 |
| |
| 作者姓名： | 王雪松 杨俊广 杨筱菡 周清雅 |
| |
| 作者单位： | 1. 同济大学 道路与交通工程教育部重点实验室, 上海 201804;2. 同济大学 数学科学学院, 上海 201804 |
| |
| 基金项目： | 上海市科学技术委员会社会发展领域项目（15DZ1204800） |
| |
| 摘 要： | 为了探究行人事故的发生机理，分析影响行人交通安全的显著因素，收集上海市中心城区263个交通分析小区（TAZ）的行人事故、道路、人口及土地利用数据，并开展行人宏观安全研究。考虑到TAZ之间存在的空间相关性，建立考虑空间相关性的贝叶斯负二项条件自回归模型，在条件自回归模型中对比分析了5种不同的空间权重矩阵，包括0~1邻接矩阵、边界长度矩阵、分析单元中心距离倒数矩阵、事故空间中心距离倒数矩阵这4种既有矩阵，以及首次引入的宏观安全建模中的分析单元中心距离多阶矩阵。结果表明：分析单元中心距离多阶矩阵的模型拟合效果和事故预测准确度均显著优于既有的4种空间权重矩阵，证明了在宏观安全建模过程中考虑研究对象交通特征（居民步行平均出行距离等）的必要性；人口数量、主干道长度、次干道长度、路网密度等因素均与行人事故呈现显著正相关，平均交叉口间距、三路交叉口比例等因素与行人事故呈显著负相关；相较于高等、低等土地利用强度，中等土地利用强度对行人事故的影响最大。
|
| 关 键 词： | 交通工程 宏观安全分析 条件自回归模型 行人事故 贝叶斯估计 空间权重矩阵 |
| 收稿时间： | 2017-07-08 |
Macro-level Pedestrian Safety Analysis Considering Spatial Correlation |
| |
| WANG Xue-song,YANG Jun-guang,YANG Xiao-han,ZHOU Qing-ya. Macro-level Pedestrian Safety Analysis Considering Spatial Correlation[J]. China Journal of Highway and Transport, 2018, 31(5): 136-142 |
| |
| Authors: | WANG Xue-song YANG Jun-guang YANG Xiao-han ZHOU Qing-ya |
| |
| Affiliation: | 1. Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University, Shanghai 201804, China;2. School of Mathematics Science, Tongji University, Shanghai 201804, China |
| |
| Abstract: | To explore the accident mechanism of pedestrian crashes and analyze the significant factors affecting pedestrian traffic safety, this research collected pedestrian crash, roadway, population, and land use data in 263 traffic analysis zones (TAZs) within the Outer Ring of Shanghai. This was done through a macro-level safety analysis to figure out the relationship between pedestrian safety and the relevant factors. Considering the spatial correlation of data, TAZ-level Bayesian negative conditional autoregressive (CAR) models were established. To characterize the existing spatial correlation more accurately and to improve the model prediction results, five different spatial weight structures (including the existing four matrices, 0-1 matrix, boundary length matrix, reciprocal matrix of the center distance of the analysis unit, and the reciprocal matrix of the crash space center distance, and the geometric centroid-distance-order matrix that was introduced for the first time in macro-level safety analysis) were established and compared using the CAR models to represent the spatial correlation between the data. The results confirmed that the geometric centroid-distance-order matrix was significantly better than the existing four matrices in terms of model fitting and crash prediction accuracy, which validated the necessity of accounting for the traffic characteristics of research subjects (i.e., the average walking distance of residents) in macro-level safety modeling. Furthermore, more pedestrian crashes occurred when the population increased. Longer arterials and minor arterials were also associated with more pedestrian crashes in each TAZ. Road density was positively correlated with pedestrian crashes, whereas average intersection spacing was negatively correlated with pedestrian crashes. The percentage of three-legged intersection was negatively correlated with pedestrian crashes. Compared to high- and low-land-use intensity, more pedestrian crashes occurred in the zones with middle-land-use intensity. |
| |
| Keywords: | traffic engineering macro-level safety analysis conditional autoregressive model pedestrian crash Bayesian estimation spatial weight matrix |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《中国公路学报》浏览原始摘要信息 |
|
点击此处可从《中国公路学报》下载全文 |
|
