Bulatov Marat Ilyasovich (MSTU «STANKIN»
Graduate student
)
Eliseeva Natalya Vladimirovna (MSTU «STANKIN»
Associate Professor, Candidate of Technical Sciences
)
| |
The article is devoted to the development of approaches to training neural networks, taking into account the specifics of urban road infrastructure management. The relevance of the study is due to the need to improve the efficiency and safety of transport systems in the context of increasing urbanization and motorization. The purpose of the work is to create a methodology for training neural networks adapted to solving problems of monitoring, forecasting and optimizing traffic in cities. To achieve this goal, methods of deep learning, transfer learning, reinforcement learning, as well as analysis of big data collected from sensors and video cameras on the roads were used. The empirical base consists of data on 5 major cities of Russia for the period 2019-2023. The results showed that the developed approaches make it possible to increase the accuracy of traffic intensity forecasting by 12-17%, reduce the average travel time by 8-11%, and reduce the number of accidents by 9-13% compared with traditional methods. The theoretical significance of the study lies in the development of a methodology for adapting neural networks to industry specifics. The practical value is associated with the possibility of implementing the results into intelligent urban transport systems. It is advisable to direct further research to expand the geography of the application of approaches and take into account related factors of urban development.
Keywords:neural networks, deep learning, transfer learning, reinforcement learning, urban road infrastructure, intelligent transport systems.
|
|
| |
|
Read the full article …
|
Citation link:
Bulatov M. I., Eliseeva N. V. DEVELOPMENT OF APPROACHES TO TRAINING NEURAL NETWORKS, TAKING INTO ACCOUNT THE SPECIFICS OF URBAN ROAD INFRASTRUCTURE MANAGEMENT // Современная наука: актуальные проблемы теории и практики. Серия: Естественные и Технические Науки. -2024. -№11/2. -С. 83-88 DOI 10.37882/2223-2966.2024.11-2.04 |
|
|
LEGAL
INFORMATION:
