Automatic tuning of parameters in a network traffic simulation model.

This study proposes a method that automatically tunes parameters in a network traffic simulation and examines its validity through applications to a real network. Quite a few simulation models have been developed to evaluate in advance not only conventional traffic policies and regulations but also recent ITS applications such as impacts of AHS (Automated Highway System), ETC (Electronic Toll Collection), TDM (Traffic Demand Management) etc. Network simulation models are thus needed for the further ITS R&D and deployment. Since the simulation models deal with dynamic traffic phenomena, they can directly consider traffic queues which are vehicles accumulated in the network. However, for their applications to real networks, experts have to spend quite a lot of efforts to calibrate the parameters inside the models. Therefore, for more frequent use of simulation models, the parameter calibration should be not only systematic but automatic. For the covering abstract see ITRD E114174.