NSF/USDOT Collaborative Proposal: Methodology for Calibration and Validation of Traffic Simulation Models

NSF/USDOT 合作提案：交通仿真模型校准和验证方法

• 批准号: 0339005
• 负责人: Moshe Ben-Akiva
• 金额: $ 7.5万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2003
• 资助国家: 美国
• 起止时间: 2003-09-15 至 2006-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0339005&HistoricalAwards=false
• 关键词: NSF; USDOT; Collaborative; Proposal; Methodology

Traffic simulation is a critical methodology for the analysis of complex transportation systems and evaluation of new technologies and concepts. Modern traffic simulation tools are a synthesis of a number of individual models, interacting in complex ways. These models capture traffic dynamics and network performance on one hand, and travel behavior on the other. Despite the importance of traffic simulation models, both in research and practice, systematic procedures for their calibration and validation are very limited. The objective of this research is to develop the basis for scientific and rigorous approaches to the calibration and validation of traffic simulation models using readily available data, e.g., loop detector counts, speeds, etc. The calibration and validation approaches to be developed recognize the aggregate nature of the available data (which makes the error associated with various components of the simulation model difficult to isolate); and the complex, large-scale nature of the applications. The research builds the required methodological foundations from diverse fields such as, stochastic optimization, statistics, traffic flow theory, and travel behavior and uses empirical analysis to provide support for theoretical developments. Important activities of the research include: Development of a rigorous modeling framework for calibration and validation.Exploration of algorithmic and computational aspects of the proposed solution approaches.Identification of appropriate statistical validation techniques, applicable to the type of data available for validation of traffic simulation models and consistent with the characteristics and properties of the measures of performance of interest.Identification of long-term research directions that will eventually lead to practical tools for calibration and validation.Empirical analysis to assess the validity of the proposed approaches. Successful completion of the research will have important broader impacts. Without appropriate validation and calibration, conclusions regarding new systems and designs can be erroneous with serious consequences in terms of deployment cost and time, and overall assessment of their value. Improvements in the accuracy of traffic simulators and the validity of the conclusions reached through simulation will result in the minimization of the risk involved in the deployment of new technologies and the development of appropriate risk mitigation strategies and better utilization of available resources. Case studies will be developed. The research results will be disseminated through scholarly journals, seminars and conferences as well as through a web page. The results will also be integrated into the curriculum at two universities.

交通仿真是分析复杂交通系统和评估新技术和新概念的关键方法。 现代交通仿真工具是许多单独模型的合成，以复杂的方式相互作用。 这些模型一方面捕获交通动态和网络性能，另一方面捕获出行行为。 尽管交通仿真模型在研究和实践中都很重要，但用于其校准和验证的系统程序非常有限。 这项研究的目的是发展的基础上，科学和严格的方法来校准和验证交通模拟模型使用现成的数据，例如，待开发的校准和验证方法认识到可用数据的聚合性质（这使得与模拟模型的各个组件相关联的误差难以隔离）;以及应用的复杂、大规模性质。 本研究从随机优化、统计学、交通流理论和出行行为等不同领域建立了所需的方法论基础，并使用实证分析为理论发展提供支持。 研究的重要活动包括：开发用于校准和验证的严格建模框架。探索拟议解决方案的算法和计算方面。识别适当的统计验证技术，适用于可用于验证交通模拟模型的数据类型，并与感兴趣的性能指标的特征和属性一致。长期的研究方向，最终将导致校准和验证的实用工具。实证分析，以评估所提出的方法的有效性。研究的成功完成将产生重要的更广泛的影响。 如果没有适当的验证和校准，关于新系统和设计的结论可能是错误的，在部署成本和时间以及对其价值的总体评估方面会产生严重后果。 提高交通模拟器的准确性和通过模拟得出的结论的有效性，将导致最大限度地减少部署新技术所涉及的风险，并制定适当的减少风险战略和更好地利用现有资源。将开展个案研究。研究结果将通过学术期刊、研讨会和会议以及网页传播。研究结果还将整合到两所大学的课程中。