Construction of bayesian trust networks for forecasting passenger flows of public transport using generative artificial intelligence

The article is devoted to the topical issue of managing congestion in public transport systems. Currently, urban population growth and increased stress on transport infrastructure are being observed, emphasizing urgency of ensuring effective management to enhance passenger comfort and reduce operating costs. In the article, recent research in this field is studied, which employs various approaches, including probabilistic models, generative adversarial networks, and optimization methods.The article proposes the approach to forecasting and modeling passenger flows in public transportation, based on the integration of Bayesian belief networks—as a tool for probabilistic modeling of dependencies between various factors—and generative artificial intelligence, in particular representatives, including Generative Adversarial Networks and Variational Autoencoders, which allow modeling and simulating potential scenarios in transport systems. This synergy of approaches enables accounting for uncertainty and dynamic changes in the transport network, as well as generating synthetic data in cases where there is a shortage of real measurements—a typical problem in transport modeling.The study is based on the analysis of empirical data from the period 2015–2018, demonstrating the practical effectiveness of the proposed model. The results confirm the approach’s ability to provide more precise load forecasting, respond more promptly to changing conditions, and form adaptive transport management strategies. The proposed tool can be useful for urban planning authorities, transport operators, and developers of intelligent transportation systems. Thus, the article has both scientific and applied value and opens prospects for further research in the field of integrating generative AI with probabilistic analysis in transport modeling. Traffic statistics for the period from 2015 to 2018 are provided.