Videre analyser af funktion til modellering af vejtrængsel i trafikmodeller

The assignment involves estimation of parameters for the congestion function (BPR-function) used in the National Traffic Model to describe the relationship between the traffic volume and the speed. Based on collected data, the ???? (alpha) and ???? (beta) parameters of the BPR-function are estimated, after which the National Traffic Model is used to calculate the effect of the estimated parameters. The collected data contains observations from Mastra and observations from Hastrid. In addition to calculations where the estimated parameters are used, sensitivity calculations are also made, in which four scenarios have been set up to examine the significance of the alpha and beta parameters for the calculations in the National Traffic Model. To estimate the alpha and beta parameters, the collected data is processed from Mastra and Hastrid, where it is compared to become wiser on which data makes the most sense to continue to work with in this project. Various analyzes of the Mastra data that is used in this project are made from 20 different selected road count stations in Denmark. The analyzes provides an overview of the traffic situation at the different sites, and based on the data processed, the alpha and beta parameters are estimated by having the parameters of the BPR-function to follow the traffic flow. By adjusting the capacity, it’s possible to make the function fit even better for the data. This project also contains analyses where vehicle data have been joined together to weather data. The analyses contains linear regressions to investigate the effect of the speed for different weather types including a new parameter estimation based on the earlier used method to compare the different estimated alpha and beta parameters to figure out if there’s any difference in the parameter estimations across the different weather types. The last analysis compares the effect of single vehicle data where the data have been aggregated into different intervals, to investigate the effect of vehicle data collected in intervals of 5, 10, 15, 30 and 60 minutes. The results from the model calculations are compared to the observations from the 20 selected data collection stations, showing if it’s possible to make an improved description of congestion, in relation to the current parameters used in the National Road Model.

本任务旨在为国家交通模型（National Traffic Model）中用于描述交通流量与车速之间关系的拥堵函数（即BPR函数（BPR-function））开展参数估计工作。基于采集得到的观测数据，将对BPR函数的α（alpha）与β（beta）参数进行估计，随后借助国家交通模型计算所得参数的影响效应。采集所得数据包含来自Mastra与Hastrid的观测记录。除使用上述估计参数开展计算外，本研究还设置了敏感性分析模块：共构建四种场景，用以检验α与β参数对国家交通模型计算结果的显著性影响。为完成α与β参数的估计，需对来自Mastra与Hastrid的采集数据进行预处理，并通过对比分析明确本项目中最具研究价值的数据集。本项目针对丹麦境内20个选定的道路计数站点的Mastra数据开展多维度分析，以厘清各站点的交通运行状况；基于处理后的数据，通过使BPR函数的参数适配交通流特征，完成α与β参数的估计。通过调整道路通行容量，可进一步优化函数与实测数据的拟合效果。本项目还包含车辆数据与气象数据的融合分析内容。此类分析采用线性回归方法探究不同天气类型下车速的影响效应；同时基于此前所用的参数估计方法开展新一轮参数估计，以对比不同天气类型下的α、β参数估计结果是否存在显著差异。最后一项分析对比了聚合至不同时间间隔的单车数据的影响效果，旨在考察采集间隔分别为5、10、15、30及60分钟的车辆数据对模型的影响程度。将模型计算所得结果与20个选定数据采集站点的观测记录进行对比，可验证基于现有国家公路模型（National Road Model）所用参数，能否实现更精准的道路交通拥堵状况描述。