A discrete-time nonlinear optimal control mechanism for monitoring dynamic signalized urban traffic networks

Abstract

Monitoring traffic operations in a detailed manner and at network level, stands for a fundamental component in optimal traffic systems control. Especially the cases of monitoring urban signalized networks, the benefits of accurate and reliable systems monitoring are essential in applying a wide range of control strategies, mechanisms and concepts, among others information provision, signal control and optimal routing. In the current paper, a discrete-time optimal control scheme is developed and tested for monitoring urban signalized networks, augmenting real-time traffic information with a dynamic traffic model. In particular, a detailed representation of network traffic based on the cell transmission model is dynamically calibrated by a quasi-Newton non-linear process able to tackle optimization cases of non-continuous variables by approximating derivatives and performing a line search. An alternative network loading assumption is tested, where the loading process covering all network links’ flows, is not relying on complex dynamic equilibrium assumptions that may infer unnecessary estimation inaccuracies, but on the combined estimation of entry flows and turning proportions at each intersection. Results from the application of the proposed framework on a realistic urban network provide encouraging evidence on its value for monitoring realistic urban traffic systems.