An adaptive Kalman filtering algorithm for the dynamic estimation and prediction of freeway origin-destination matrices

Shou-Ren Hu, Purdue University

Abstract

The purpose of this research is to develop a dynamic model for on-line estimation and prediction of freeway origin-destination (OD) demands. Time-dependent OD matrices are essential inputs for Advanced Traffic Management and Information Systems (ATMIS). With the information contained in time-varying OD matrices, it is possible to project travel demands up to a time horizon of interest and predetermine optimal control and routing policies that achieve some desirable system objectives. Therefore, an effective model for the dynamic estimation and prediction of freeway OD matrices is crucial to on-line traffic management systems. In the present study, an adaptive Kalman Filtering (KF) algorithm with route switching consideration has been developed. The major innovation of this research is the inclusion of a control equation within the Kalman Filtering algorithm, which explicitly captures the time-varying effects of traffic information on travel demand distributions through a behavioral model of route switching. In addition, the use of a traffic simulator as a travel time predictor to predict time-varying travel time model parameters has been shown to be promising for the dynamic estimation of freeway OD matrices. With the real-time feedback feature, the proposed adaptive KF estimator is applicable to on-line traffic management systems. The proposed methodology was evaluated through simulation experiments using a hypothetical network and historical OD data set. The test results demonstrate the capability of the adaptive Kalman Filtering algorithm for the dynamic estimation and prediction of freeway OD demands. Moreover, the issues of using time-varying model parameters, the effect of traffic information, and adaptive estimation have been addressed and tested in this research.

Degree

Ph.D.

Advisors

Madanat, Purdue University.

Subject Area

Civil engineering|Electrical engineering|Transportation

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS