Applications of traffic simulators to problems such as automatic incident detection require very accurate flow reconstruction, e.g. in order to distinguish abnormal situations from systematic discrepancies between model and reality. The quantitative and even qualitative performance depend critically on correct calibration of phenomenological constants in the underlying traffic model. Simulation of heavy traffic conditions is of crucial importance in many applications but pose a severe challenge for traffic simulation. The authors introduce a useful measure of goodness-to-fit, report the results of a calibration performed with the recently developed mesoscopic traffic simulator ANIMAL (ANIsotropic Mesoscopic ALgorithm) under conditions of heavy traffic, and examine the performance with this optimal calibration. (A)
Samenvatting