Problem statement

The simulation of traffic flow on freeways - including the effect of freeway control systems - is a key tool for optimizing traffic control. However, this potential is still too rarely used, as the simulation of complex control algorithms is associated with considerable effort. Outdated or incomplete documentation makes it difficult to reconstruct the control system accurately and requires time-consuming adjustments. Nevertheless, efficient mapping of the algorithms is essential in order to enable realistic and meaningful simulations and thus improve traffic control in the long term.

Objective

KISA aims to reconstruct control algorithms using machine learning from existing and easily accessible traffic and switching data. This allows missing information on local specifications to be avoided. These digital twins of real route control systems enable realistic traffic flow simulations.

Methods

Traffic and switching data from freeway control systems serve as the basis for training neural networks and replicating the underlying control algorithms. Classical optimization methods, such as parameter estimation, are also used. Data from traffic flow simulations support the systematic development of the AI models.