Modeling origin-destination uncertainty using network sensor and survey data and new approaches to robust control |
| |
Institution: | 1. Department of Physics, University of Crete, P.O. Box 2208, 71003 Heraklion, Greece;2. Microelectronics Research Group, IESL-FORTH, P.O. Box 1385, 71110 Heraklion, Greece;1. Advanced Robotics and Automation System (ARAS), Industrial Control Center of Excellence, Faculty of Electrical Engineering,;2. Advanced Robotics and Automation System (ARAS), Industrial Control Center of Excellence, Faculty of Electrical Engineering, K. N. Toosi University of Technology, Tehran, Iran;3. Advanced Robotics and Automation System (ARAS), Industrial Control Center of Excellence, Faculty of Electrical Engineering, K. N. Toosi University of Technology, Tehran, Iran |
| |
Abstract: | This study develops new methods for network assessment and control by taking explicit account of demand variability and uncertainty using partial sensor and survey data while imposing equilibrium conditions during the data collection phase. The methods consist of rules for generating possible origin–destination (OD) matrices and the calculation of average and quantile network costs. The assessment methodology leads to improved decision-making in transport planning and operations and is used to develop management and control strategies that result in more robust network performance. Specific contributions in this work consist of: (a) Characterization of OD demand variability, specifically with or without equilibrium assumptions during data collection; (b) exhibiting the highly disconnected nature of OD space demonstrating that many current approaches to the problem of optimal control may be computationally intractable; (c) development of feasible Monte Carlo procedures for the generation of possible OD matrices used in an assessment of network performance; and (d) calculation of robust network controls, with state-of-the-art cost estimation, for the following strategies: Bayes, p-quantile and NBNQ (near-Bayes near-Quantile). All strategies involve the simultaneous calculation of controls and equilibrium conditions. A numerical example for a moderate sized network is presented where it is shown that robust controls can provide approx. 20% cost reduction. |
| |
Keywords: | OD uncertainty Robust optimization Traffic signal control Near-Bayes near-quantile strategy Disconnected OD space |
本文献已被 ScienceDirect 等数据库收录! |
|