Enhanced Kalman Filtering algorithm for dynamic freeway OD matrix estimation and prediction

This paper presents an enhanced Kalman Filtering algorithm for the dynamic estimation and prediction of freeway OD matrices. The effects of traffic congestion and traffic diversion information on the OD distribution patterns are explicitly captured through a behavioral model of route switching. In view of time-varying nature of traffic movements, the proposed algorithm updates the model parameters by using on-line traffic measurements. Preliminary simulation results demonstrate the importance of using time-dependent model parameters and accounting for the effect of traffic information in the estimation of dynamic OD demands.