Back to Search
Start Over
Training and Applying Artificial Neural Networks in Traffic Light Control: Improving the Management and Safety of Road Traffic in Tyumen (Russia).
- Source :
- Automotive Experiences; 2023, Vol. 6 Issue 3, p528-550, 23p
- Publication Year :
- 2023
-
Abstract
- The article presents the initial experience (spring-summer 2023) of using artificial neural networks (ANN) in improving traffic management in the large Russian city of Tyumen. Using the example of one of the intersections of the city's road network, it is shown how much transport delays are reduced when the duration of the traffic light cycle phases is quickly adjusted to the actual traffic intensity when compared with the usual previously used traffic light predictive mode. For the specific intersection of Odesskaya and Kotovskogo streets in Tyumen, considered in this article, the traffic light control mode using an ANN can significantly (by 20.6 ... 22.4%) reduce the average delay time of vehicles. It is also important that the reduction in traffic delays, which is possible with the regulation of traffic using ANN, helps to reduce stress for road users and improve road safety. The article presents historical data illustrating the dynamics of changes in the field of traffic management and road safety in Tyumen. This information well confirms the thesis about the dialectic of systemic development and the need for a gradual increase in the intellectual component in traffic management in large cities. The Applications (Appendix A and Appendix B) present the code of the auxiliary procedures and functions module and the code of the main data collection module used to optimize the traffic light control mode at the experimental intersection of the Tyumen road network. The main conclusion of the study is that use of an ANN allows taking into account a much larger number of factors and optimizing the control of the entire object, consisting of several intersections, which is not achievable using predictive modes and local adaptive control. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 26156202
- Volume :
- 6
- Issue :
- 3
- Database :
- Complementary Index
- Journal :
- Automotive Experiences
- Publication Type :
- Academic Journal
- Accession number :
- 175439829
- Full Text :
- https://doi.org/10.31603/ae.10025