A new dynamic OD estimation method in ITS based on fuzzy DTA model and Kalman filtering

In this paper, dynamic OD matrix estimation and prediction from real-time link counts based on fuzzy dynamic traffic assignment (FDTA) and Kalman Filtering was proposed. At first, the state-space model of dynamic OD matrix estimation and prediction was established. To obtain the key assignment matrix in measurement equation of the state-space model, fuzzy dynamic traffic assignment (FDTA) was applied based on C-LOGIT model and fuzzy shortest path (FSP) algorithm in which fuzzy path travel time (FPTT) was constructed from real-time link counts. Then Kalman Filtering algorithm iterated with FDTA was used for dynamic OD estimation and prediction. Simulation test shows that compared with K. Ashok's Kalman Filter model, the mean estimation error and the mean square error of FDTA-based Kalman Filter model are improved greatly, which exactly means more precision and better dynamic characteristics of dynamic OD estimation and prediction could be obtained.