University of Twente Student Theses


Gravity model parameter calibration for large scale strategic transport models.

Pots, M.H. (2018) Gravity model parameter calibration for large scale strategic transport models.

[img] PDF
Abstract:This study considers the calibration of the lognormal cost function parameters within the simultaneous gravity model for large scale strategic transport models. The parameters are calibrated based on observed trip length distributions. A new calibration approach is investigated that proposes a triproportional fitting procedure. The procedure solves the gravity model equations together with modal split constraints. As a consequence the calibration is simplified and its running time is considerably reduced.The calibration problem for this new approach then is formulated as a bilevel optimization problem. In this formulation the outer optimization task is to choose the parameters and the inner optimization task is to maximize entropy subject to trip end and modal split constraints. A BFGS Quasi Newton method was developed to solve the outer optimization task and was compared with the currently used hillclimbing algorithm at Goudappel Coffeng. This new method converges more reliably to the local minimum albeit much slower. Further two methods to compute the gradient are considered. The first one a finite differences approach. The second method an analytical adjoint method and that was observed to be significantly faster for smaller to medium scale models, however occasionally suffered from a singular linear system.
Item Type:Essay (Master)
Goudappel Coffeng, Deventer, Netherlands
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:55 traffic technology, transport technology
Programme:Applied Mathematics MSc (60348)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page