EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Breaking Is Easier than Building: Improving Public Transport Routes to Enhance Travel-Time Connectivity Using a Neural Evolutionary Approach

Aleksandr Morozov, Ruslan Kozlyak, Yana Zorina, Sergey A. Mityagin

Smart Cities · 07.11.2026 · 9 (11) · 191

DOI: 10.3390/smartcities9110191 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Smart Cities, online erschienen am 07.11.2026. Verfasst von Aleksandr Morozov und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: The Ministry of Economic Development of the Russian Federation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Improving an existing public transport route network is a challenging optimization problem, since service quality and operational efficiency must be improved without excessive changes to the established network. Most public transport network improvement methods modify route networks without considering the preservation of existing routes. At the same time, recent applications of graph neural networks combined with evolutionary algorithms to transport network design have demonstrated that learned heuristics can effectively guide the search for improved route networks. We aim to extend this approach by training the neural network to improve an existing network rather than to construct one from scratch. In this study, the neural evolutionary approach is applied to the improvement of existing public transport route networks with a soft constraint on the degree of route adjustment. Experiments on the Mandl and Mumford benchmark instances show that the learned route-editing heuristic achieved lower mean passenger-oriented costs than the original neural approach. The proposed methods are evaluated in a case study of Yekaterinburg, Russia, resulting in a solution that simultaneously reduces weighted median connectivity time by 69.6% and total route travel time by 26.3% relative to the initial network. This solution has the lowest total route travel time among the evaluated configurations. These case-study gains are conditional on an expanded candidate-stop set that includes locations not served by the initial network.

Förderung

  • The Ministry of Economic Development of the Russian Federation

Zitieren

APA

Morozov, A., Kozlyak, R., Zorina, Y., & Mityagin, S. A. (2026). Breaking Is Easier than Building: Improving Public Transport Routes to Enhance Travel-Time Connectivity Using a Neural Evolutionary Approach. Smart Cities, 9(11), 191. https://doi.org/10.3390/smartcities9110191

Vancouver

Morozov A, Kozlyak R, Zorina Y, Mityagin SA. Breaking Is Easier than Building: Improving Public Transport Routes to Enhance Travel-Time Connectivity Using a Neural Evolutionary Approach. Smart Cities. 2026;9(11):191. doi:10.3390/smartcities9110191

BibTeX
@article{Morozov2026,
  author = {Morozov, Aleksandr and Kozlyak, Ruslan and Zorina, Yana and Mityagin, Sergey A.},
  title = {Breaking Is Easier than Building: Improving Public Transport Routes to Enhance Travel-Time Connectivity Using a Neural Evolutionary Approach},
  journal = {Smart Cities},
  year = {2026},
  volume = {9},
  number = {11},
  pages = {191},
  doi = {10.3390/smartcities9110191}
}
Kostenlos starten →