Description
The use of Agent-Based and Activity-Based modeling in transportation is quickly rising due to the capability of addressing more complex applications, such as the evaluation of emerging mobility patterns or of disrupting mobility technologies. Indeed, large-scale disaggregate demand modeling rooted in behavioral theory is more relevant than ever due to disruptive trends such as remote working and automation. Still, a wide adoption of these models is not materializing due to the high complexity and computational needs. While the latter are addressed by the ever-increasing computational capacities, activity-based modeling focused on behavioral theory may involve hundreds of parameters and requires a detailed socio-economical characterization of any case study. This paper tackles these issues by presenting a Bayesian Optimization algorithm designed to automatethe calibration process of the behavioral parameters. A surrogate model is developed and tested to calibrate 477 behavioral parameters for the SimMobility MT software. The achieved results are reported and compared against the baseline distributions for the city of Tallinn, Estonia. A satisfactory match is achieved in most of the indicators: the error for the overall number of trips is equal
to 4%, the average error in the OD matrix is 15.92 vehicles per day, and the work or educational trips replicate the spatial anchor points distribution.
| Aikajakso | 11 tammik. 2023 |
|---|---|
| Tapahtuman otsikko | Transportation Research Board Annual Meeting |
| Tapahtuman tyyppi | Conference |
| Konferenssinumero | 102 |
| Sijainti | Washington, Yhdysvallat, District of ColumbiaNäytä kartalla |
| Tunnustuksen arvo | International |