Regional Transfer Flow Ratio Estimation in a Multi-Reservoir Traffic System

The management of inflows and outflows between reservoirs in large-scale traffic flow models has been a prominent research topic in recent years. Prior studies have utilized either regional path flow distribution methods along with Origin-Destination (OD) information or real-time measurements from loop detectors to determine regional transfer flow ratios. This paper presents a novel approach for estimating outflow ratios across reservoirs using Floating Car Data (FCD). The method involves the estimation of turning rates at specific intersections and queue lengths at different links to derive regional transfer flow ratios. The performance of two estimation methods and various penetration rates are evaluated using synthetic data. The first method directly uses FCD for estimation, while the second method incorporates a smoothing technique with estimation to minimize errors. The results demonstrate that the accuracy of estimation is enhanced with the smoothing technique, particularly when the mean penetration rate is below 10%. Furthermore, it is shown that having greater penetration rates than 10% can not noticeably increase the accuracy. The validation of the proposed method in the test bed of this study suggests that having an average of 10% probe vehicles within the network is a trustworthy quantity for identifying regional transfer flow ratios.