National Technical University of Athens (NTUA)
Towards data-driven microscopic traffic simulation models
Abstract
dc:descriptionThe objective of this research is to develop more accurate, robust and reliable microscopic models. Anintegrated methodological framework based on non–parametric approaches is proposed for estimationof data–driven microscopic traffic simulation models. The methodology is implemented using differentma-chine learning techniques such as clustering, classification, locally weighted regression, splinefitting, Gaussian processes, Kernel support vector machines and neural networks. The methodology isdemonstrated using real trajectory data from three different sources and specifically an experimentfrom Naples, NGSIM data and non–lane disciplinary trajectory data from India. The focus is given oncar–following models and Gipps’ model, one of the most extensively used car–following models, iscalibrated against the same data in order to be used as a reference benchmark. Many parametersaffect driving behavior and it is explored how the performance of the models is improved by includingmore explanatory variables. Then, a practical and simple approach is developed and motivated for theonline calibration of microscopic traffic simulation models, which considers dynamic parameters forindividual drivers, in time and space. The model adapts to driving behavior in a rolling horizon andleads to less than 10% error in speed prediction even for ten steps into the future. This research alsoexamines the feasibility and the benefits of using data–driven models on mixed traffic trajectory data,including non–lane discipline and heterogeneity in vehicle types, common characteristics in cities indeveloping countries. Although typical car–following models are theoretically justified, data–drivenapproaches are more flexible and allow the easy incorporation of additional information to the processof speed estimation. The results indicate that data–driven models could ensure reliability andimprovement in estimation of microscopic models.
Degree
thesis:*- Grantor dc:publisher
- National Technical University of Athens (NTUA)
- Year dc:date
- 2018
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Papathanasopoulou, Vasileia
- Παπαθανασοπούλου, Βασιλεία
Subjects
dc:subject × 20- Μοντέλα βασισμένα σε δεδομένα
- Μικροσκοπικά μοντέλα κυκλοφοριακής προσομοίωσης
- Μοντελοποίηση κυκλοφορίας
- Μηχανική μάθηση
- Δυναμική βαθμονόμηση
- Δδομένα NGSIM
- Μη τήρηση λωρίδων κυκλοφορίας
- Data-driven models
- Microscopic traffic simulation models
- Traffic modeling
- Machine learning
- On-line calibration
- NGSIM data
- Non-lane discipline
- Επιστήμες Μηχανικού και Τεχνολογία
- Επιστήμη Πολιτικού Μηχανικού
- Άλλες Επιστήμες Μηχανικού και Τεχνολογίες
- Engineering and Technology
- Civil Engineering
- Other Engineering and Technologies
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/44979