Incident indicators for freeway traffic flow models

Artikel i vetenskaplig tidskrift, 2022

Developed in this paper is a traffic flow model parametrised to describe abnormal traffic behaviour. In large traffic networks, the immediate detection and categorisation of traffic incidents/accidents is of capital importance to avoid breakdowns, further accidents. First, this claims for traffic flow models
capable to capture abnormal traffic condition like accidents. Second, by means of proper real-time estimation technique, observing accident related parameters, one may even categorize the severity of accidents. Hence, in this paper, we suggest to modify the nominal Aw-Rascle (AR) traffic model by
a proper incident related parametrisation. The proposed Incident Traffic Flow model (ITF) is defined by introducing the incident parameters modifying the anticipation and the dynamic speed relaxation terms in the speed equation of the AR model. These modifications is proven to have physical meaning.
Furthermore, the characteristic properties of the ITF model is discussed in the paper. A multi stage numerical scheme is suggested to discretise in space and time the resulting non-homogeneous system of PDEs. The resulting systems of ODE is then combined with receding horizon estimation methods
to reconstruct the incident parameters. Finally, the viability of the suggested incident parametrisation is validated in a simulation environment.