Investigation of a stochastic disaggregation model.

Dynamic coupling of a microscopic (or mesoscopic) highway traffic simulator to boundary conditions known only at an aggregated level requires generation of individual vehicle data, a process referred to as disaggregation. In order for a true dynamic coupling to take place, this individual vehicle data obviously cannot be purely synthetic, but must faithfully reproduce the information in the macroscopic data. Ideally, the statistical distributions of gaps and times to collision (TTC) as well as velocity correlations and fluctuation levels of the synthetic time series data generated by the disaggregator should also correspond to those of real data. Incorrect distributions couldd cause errors in the representation of microscopic phenomena depending on these distributions such as merging and passing. This paper reports on a stochastic disaggregator that combines aggregated (macroscopic) data with historical information to produce individual vehicle data. This disaggregated data is available directly at the interface to a microscopic (or mesoscopic) traffic simulator in the form of appropriate time series. For the covering abstract see ITRD E114174.