Calibrating Activity-Based Models with External Origin-Destination Information
Calibrating Activity-Based Models with External Origin-Destination Information
复制标题
DOI：
10.3141/2175-12
复制
发表时间：
2010-12
影响因子：
1.7
通讯作者：
M. Cools;E. Moons;G. Wets
中科院分区：
文献类型：
--
作者：
M. Cools;E. Moons;G. Wets
许多从业者质疑基于活动的模型在流量计数复制方面相对于传统四步模型的优势。本文重点介绍了一个框架，该框架将一般的出行需求模型和特别是基于活动的模型与交通流量主动联系起来。提出了两种用流量计数来校准基于活动的模型的方法——间接方法和直接方法。间接方法试图将基于流量计数分析的结果纳入基于活动的模型的组成部分。直接方法校准出行需求模型的参数，使模型完美地（准）复制观察到的交通量。提供了一个实际例子来说明直接方法。本实例的研究区域是比利时哈瑟尔特市，这座城市拥有约 70,000 名居民及其周边城市。实际示例揭示了校准基于活动的模型没有一条成功之路，但在微调基于活动的模型方面存在不同的选择。认识到一些悬而未决的问题和进一步研究的途径非常重要。首先，假设流量计数完全正确并不总是合适的。建立一些信念结构可能会提高基于活动的模型的响应能力。此外，优化估计和观察到的屏幕线计数之间的对应关系的起点-目的地矩阵校准可能会对与其他测量值（例如车辆行驶里程）的对应关系产生负面影响。总之，多目标校准方法的制定是进一步研究的关键挑战。
Many practitioners question the advantages of activity-based models over conventional four-step models in regard to replication of traffic counts. This paper highlights a framework that actively links travel demand models in general and activity-based models in particular with traffic counts. Two approaches are presented that calibrate activity-based models with traffic count–an indirect and a direct approach. The indirect approach tries to incorporate findings based on the analysis of traffic counts into the components of the activity-based models. The direct approach calibrates the parameters of the travel demand model in such a way that the model replicates the observed traffic counts (quasi-) perfectly. A practical example is provided to illustrate the direct approach. The study area for this practical example is Hasselt, Belgium, a city of about 70,000 residents, and its surrounding municipalities. The practical examples revealed not a single roadway to success in calibrating activity-based models, but different options exist in fine-tuning the activity-based model. It is important to recognize some open issues and avenues for further research. First, it is not always appropriate to assume that traffic counts are completely correct. Setting up some belief structure might increase the responsiveness of the activity-based model. In addition, the origin-destination matrix calibration that optimizes the correspondence between estimated and observed screen-line counts could negatively affect the correspondence to other measures, such as vehicle miles traveled. To conclude, the formulation of a multiobjective calibration method is a key challenge for further research.