An integrated approach to apdative anticipatory traffic control and parameter estimation

This paper develops an integrated approach to combine adaptive anticipatory network traffic control and model parameter estimation. The main objective is to complete an iterative control optimization method that has been proposed to drive the traffic system towards its true optimal operating point despite model uncertainty [1], in which the underlying network equilibrium model is imperfectly calibrated. In this integrated framework, two objectives are fulfilled, i.e. designing optimal reality-tracking control, and identifying a better model description of the real system at its current operating condition. The interaction among control setting update, model bias update and model parameter update, as well as its influence on the performance of the integrated framework is analyzed. Numerical tests verify the effectiveness of the proposed integrated approach in jointly performing adaptive signal control and route choice behavioral parameter estimation. It is demonstrated that the parameter estimation reduces the discrepancy between model and measured reality, while at the same time guiding the traffic network towards optimal anticipatory control despite mismatch between reality and the model used to anticipate the response of the users.