Kalman filtering based dynamic OD matrix estimation and prediction for traffic systems

In this paper, a state space model is proposed so that the dynamic OD matrix can be estimated through the surveillance of flows and traveling time on links in a traffic network. To eliminate the influence of slow time-variant parameters, a Recursive Least Square (RLS) algorithm is introduced to identify the system matrix online. Moreover, an analytical formula to calculate the key assignment matrix is presented. With the sequential Kalman Filtering method, the fast and real-time OD estimation and prediction algorithm is established. The algorithm is proven to be very effective and efficient with simulation test.