Development and Evaluation of Online Estimation Methods for a Feedback-Based Freeway Ramp Metering Strategy

The critical density of a freeway link is subject to changes over time owing to such circumstances as environmental conditions (snow, rain, etc.) and traffic incidents. Because of the critical density’s impacts on the performance of some ramp metering strategies that make use of it as a threshold value for control action, it is necessary to trace the real value of the critical density. Hence, methodologies for the online estimation of critical density using the extended Kalman filter (EKF) and Kalman filter (KF) are proposed in this paper. Basically, critical density is chosen as the state variable to be determined using the system output, namely the measurement of traffic flow on the downstream freeway link. These methods are compared with a similar methodology by Smaragdis et al. (2004) using the same traffic conditions implemented in a macroscopic simulation model based on MATLAB. Then, these proposed methods are evaluated in microscopic simulation environments using PARAMICS. In tracing timedependent changes of real critical occupancy effectively, the results indicate that both methods (using KF and EKF) provide better performance compared with the case where constant critical occupancy is used for all the scenarios tested. Our paper contributes to state of the art by using a novel on-line parameter estimation techniques combined with two different feedback based ramp metering strategies namely ALINEA and MIXCROS and then by testing this new approach in a highly stochastic microscopic simulation environment that can be considered a close representation of real-world conditions. TRB 2006 Annual Meeting CD-ROM Paper revised from original submittal. Ozbay, K., I. Yasar, and P. Kachroo 2