Feedback Traffic Control System Based on Unscented Kalman Filter

点击翻译按钮获取中文摘要

This study aims to propose a new approach for dynamic estimation of origin-destination (OD) matrices based on Unscented Kalman Filter (UKF). In the approach, the nonlinearity between OD parameters and measurements is retained by dynamic traffic simulation model.This study focuses on the data scarcity problem ; (1) reduction of unknowns by replacing OD flows with origin flows and destination flows, (2) comparison of simulation models to map OD flows and measurement variables more precisely, and (3) consideration of additional measurement data, including spot speed data and probe vehicle data.Numerical results support the potential of the proposed approach from 1) the accuracy of the estimated OD flows, 2) the effectiveness of the joint estimation of traffic distribution and traffic assignment, and the contribution of spot speed data.