Reconstructive method of missing data for location-specific detector considering spatio-temporal relationship
-
摘要: 应用相关性理论, 研究了交通流数据中缺失值与其他数据的相关性, 对与缺失值不同相关性的数据给予不同的权重值, 提出了基于交通流时空相关权重的重构算法, 并以北京市二环快速路为研究对象, 运用VISSIM仿真软件建立仿真模型, 利用仿真数据对新算法和现有算法进行了对比分析。研究结果表明: 在连续缺失1~10个数据时, 模型1的重构值与仿真值平均相对误差最大仅为1.8766%, 一般情况下, 平均相对误差均在1.0000%以下, 可见, 模型1算法优于现有的重构算法。Abstract: The relationship between the missing data and other data of traffic flow was studied by using correlation theory. A reconstructive method based on spatio-temporal relative weight values was put forward when other data had different correlations with missing data and were given different spatio-temporal relative weight values. The Beijing Expressway of Second Ring Road was selected as test road, its simulation models were established by using VISSIM, and the new method and existing methods were compared by using simulation data. Analysis result shows that the maximum average relative error between the reconstruction values and simulation values of model 1 is only 1.876 6% and below 1.000 0% in general when the numbers of continuous missing data are from 1 to 10, so the method of model 1 is better than existing reconstructive methods.
-
图 1 相邻时刻相关系数计算
Figure 1. Calculation of correlation coefficients for adjoining time data
图 2 相邻车道相关系数计算
Figure 2. Calculation of correlation coefficients for adjoining lane data
图 3 缺失数据时空相关数据及其相关系数
Figure 3. Spatio-temporal correlative data and correlation coefficients of missing data
图 5 模型1与模型3、4、5、6平均相对误差比较
Figure 5. Comparison of average relative errors among model 1 and model 3, 4, 5, 6
图 6 模型1和模型2平均相对误差比较
Figure 6. Comparison of average relative errors between model 1 and model 2
表 1 重构值与仿真值平均相对误差
Table 1. Average relative errors of reconstructive data and simulation data
% 连续缺失个数 模型1 模型2 模型3 模型4 模型5 模型6 1 1.876 6 2.384 8 5.846 2 4.570 5 4.064 1 3.667 9 2 0.627 6 0.734 4 2.384 6 1.500 0 1.262 8 1.121 8 3 0.371 1 0.510 0 2.038 5 1.179 5 0.829 1 0.760 3 4 0.017 7 0.038 9 1.384 6 2.968 0 0.134 6 0.057 4 5 0.012 0 0.024 7 0.030 8 1.731 2 0.079 5 0.148 7 6 0.019 6 0.035 1 0.346 2 1.388 0 0.040 6 0.120 7 7 0.018 8 0.044 6 0.098 9 2.645 7 0.190 5 0.079 7 8 0.057 1 0.083 2 0.894 2 4.133 7 0.267 6 0.155 3 9 0.023 7 0.023 6 0.205 1 3.171 5 0.068 4 0.042 7 10 0.082 0 0.111 4 1.315 4 2.959 5 0.160 3 0.184 7 
下载: 导出CSV
-
[1] 王晓原, 吴芳, 朴基男. 基于粗集理论的交通流丢失数据补齐方法[J]. 交通运输工程学报, 2008, 8(5): 91-94. http://transport.chd.edu.cn/article/id/200805018WANG Xiao-yuan, WU Fang, PIAO Ji-nan. Filling method of missing datafor trafficflowbased onroughset theory[J]. Journal of Traffic and Transportation Engineering, 2008, 8(5): 91-94. (in Chinese) http://transport.chd.edu.cn/article/id/200805018 [2] JOU YJ, WEN Y H, LEE T T, et al. Missing data treatment on travel time estimation for ATIS[C]∥IEEE. IEEE Inter-national Conference on Systems, Man and Cybernetics. Tai wan: IEEE, 2003: 102-107. [3] 金盛. 环形线圈检测器交通数据预处理方法研究[D]. 长春: 吉林大学, 2007.JINSheng. Research on preprocessing methods of loop detector data[D]. Changchun: Jilin University, 2007. (in Chinese) [4] 金逸文. 城市快速路交通流数据修复方法研究[D]. 上海: 上海交通大学, 2008.JI N Yi-wen. Research on repair methods for urban express-way traffic flowdata[D]. Shanghai: Shanghai Jiaotong Uni-versity, 2008. (in Chinese) [5] CLARK S, GRANT-MULLER S. CHEN Hai-bo. Cleaning of matched license plate data[J]. Transportation Research Record, 2002(1804): 1-7. [6] VAN LI NT J W C, VAN ZUYLEN H J. Monitoring and predicting freeway travel time reliability-using width and skewof the day-to-day travel time distribution[J]. Transpor-tation Research Record, 2005(1917): 54-62. [7] VAN LI NT J W C. Reliable travel time prediction for free-ways[D]. Delft: Delft University of Technology, 2004. [8] WANG Yi-bing, PAPAGEORGIOU M. Real-time freeway traffic state estimation based on extended Kal man filter: a general approach[J]. Transportation Research Part B: Meth-odological, 2005, 39(2): 141-167. [9] 韩卫国, 王劲峰, 胡建军. 交通流量数据缺失值的插补方法[J]. 交通与计算机, 2005, 23(1): 39-42. https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS200501009.htmHAN Wei-guo, WANG Jin-feng, HU Jian-jun. Imputation methods of missing values in traffic flow data[J]. Computer and Communications, 2005, 23(1): 39-42. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-JTJS200501009.htm [10] 杨庆芳. 先进的交通管理系统关键理论与方法研究[D]. 长春: 吉林大学, 2004.YANG Qing-fang. Study on critical theories and methods about advanced traffic management system(ATMS)[D]. Changchun: Jilin University, 2004. (in Chinese) -
点击查看大图
计量
- 文章访问数: 635
- HTML全文浏览量: 81
- PDF下载量: 547
- 被引次数: 0
