CAREER: Efficient and Robust On-Line Control of Large-Scale Dynamic Traffic Systems with Information Systems

摘要提案：CMS 97-02612 PI: Srinivas Peeta，普渡大学基于信息系统的大规模动态交通系统的高效鲁棒在线控制本研究以具有容量约束和系统输入中不同程度不确定性的大规模人为动态系统为例，研究了大规模交通系统的性能优化和高效在线鲁棒控制。在这种情况下，流必须从一个节点在线路由到另一个节点，而不需要对系统的未来条目/需求提供完整的先验信息。目前缺乏如此准确和详细的交通系统状态实时数据，阻碍了对控制系统性能的动态现象的清晰理解，限制了大型交通系统可行和有效的实时控制策略的发展。本研究将利用非线性自适应鲁棒控制理论和神经网络来开发在线控制策略的新领域，以寻求在特定未来需求和网络状态条件下大规模交通系统的性能优化。这项工作将包括理论、模拟和操作实验。在印第安纳州交通运输部和休斯交通管理系统的合作下，从最近安装的先进交通管理系统（ATMS）中获得的实时数据将用于验证和评估拟议的模型和算法，从而提高现有道路网络的性能效率。教育目标是通过教学和实验，系统地弥合交通领域信息系统不断发展的技术进步与现有课程缺失之间的差距，特别是在先进传感器系统、在线数据提取和处理技术、鲁棒在线控制理论和高性能计算技术等领域。本科学生将参与实时数据访问和处理，并且强调专业发展和扩展课程开发的项目包括在配套教育计划中。＊＊＊

Abstract Proposal: CMS 97-02612 PI: Srinivas Peeta, Purdue University "Career: Efficient and Robust On-Line Control of Large- Scale Dynamic Traffic Systems with Information Systems" " This research addresses the performance optimization and the efficient on-line robust control of large-scale traffic systems as a substantive example of large-scale man-made dynamic systems with capacity constraints and different degrees of uncertainties in system inputs, in which flows must be routed on-line from one node to another without complete a priori information on future entries/demands to the system. The current lack of such accurate and detailed real-time data on the state of the traffic system has precluded a clear understanding of the dynamic phenomena that govern system performance, limiting the development of feasible and effective real-time control strategies for large traffic systems. This research will use nonlinear adaptive and robust control theory and neural networks to develop a new realm of on-line control strategies that seek the performance optimization of large-scale traffic systems under certain future demands and network state conditions. The work will involve a blend of theory, simulation, and operational experiments. Real-time data obtained from a recently installed advanced traffic management system (ATMS) on a major interstate corridor in an on-going collaborative effort with the Indiana Dept. of Transportation and Hughes Transportation Management Systems will be used to validate and evaluate the proposed models and algorithms, leading to an enhanced performance efficiency of existing road networks. The education objectives are to systematically bridge the gap between evolving technological advances in information systems in the transportation arena and their absence from existing curriculum through teaching and experimentation, specifically in the areas of advanced sensor systems, on- line data extraction and processing techniques, robust on- line control theory, and high performance computational techniques. Undergraduate students will be involved with real-time data accessing and processing, and programs accenting professional development and extension course development are included in the companion educational plan. ***