Optimizing the Efficiency and Equity of Traffic Flow: Development of a Control Strategy for Freeway Corridors Using Dynamic Bayesian Decision Networks

Ramp metering has the potential to improve the safety and efficiency of existing transportation infrastructure. By controlling the flow of vehicles onto the freeway, the extent of mainline congestion can be reduced, resulting in a system-wide reduction in travel time even with the introduction of ramp delays. However, despite the potential benefits of ramp meters, their use remains controversial, in part due to equity concerns: while some drivers must wait at entrance ramps, others are allowed to access the freeway with little or no delay.;In this research, a new ramp control algorithm was developed which balances efficiency and equity objectives. To capture the uncertainty inherent in traffic systems, the algorithm has been implemented as a dynamic Bayesian decision network, and incorporates a probabilistic model of freeway flow. This thesis describes the development of the algorithm, its key features, and the results of simulation tests which demonstrate its effectiveness compared to more traditional efficiency-maximizing algorithms such as ALINEA.;Comparison with the ALINEA algorithm suggests that the new algorithm is able to achieve similar congestion benefits when operated on an efficiency basis alone. When equity considerations are also included, the system operates less efficiently, but with ramp delays that are more fairly distributed amongst drivers. Such a system has the potential to enhance public support for ramp metering -- support which is essential for the successful deployment of ramp metering in a Canadian context.;To effectively balance competing objectives, new algorithms are needed --- algorithms which leverage new technologies, bringing together the latest research from fields as diverse as artificial intelligence and industrial control. It is not enough to measure the success of an algorithm by how well it improves traffic flow. It must also meet the needs of system users, who may value ramp delays differently than delays due to freeway congestion, and who may be willing to trade-off some of the potential gains in operational performance for a system that operates more fairly.