TRAFFIC DEMAND ESTIMATION BASED ON OFFLINE TRAINED ARTIFICIAL NEURAL NETWORK
Abstract
The current paper focuses on traffic demand estimation problem. Artificial Neural Network (ANN) proposed as a prediction model. The given problem formulated as a supervised learning classification task. The dataset for model training and validation consists of synthetic data that was generated by using simulator. The results of experiments show training accuracy = 82.2 %. The evaluation of the test set gives 80.03 % accuracy. Finally, well-trained estimator of traffic flow is obtained.
About the Authors
G. Tolebi
Kazakh-British technical university
Kazakhstan
Kazakhstan
D. Kurmankhojayev
Kazakh-British technical university
Kazakhstan
Kazakhstan
References
1. Du S., Li R., Gong X., Hong S., A Hybrid Method for Traffic Flow Forecasting Using Multimodal Deep learning. Machine Learning. Cornelle University, 2019. https://arxiv.org/pdf/1803.02099.pdf
2. Bernico M., Deep Learning Quick Reference. Packt Publishing, 2018.
3. SUMO - Simulation of Urban Mobility. Institute of Transportation Systems. Available at: http://sumo.dlr.de/wiki/SUMO
4. Kurmankhojayev D., Tolebi G., Analysis of the traffic flow modeling systems. Herald of the Kazakh-British technical university, № 4 (47), 2018. pp. 31-36.
5. Kurmankhojayev D., Suleymenov N., Tolebi G., Online model-free adaptive traffic signal controller for an isolated intersection. 2017 International M ulti-Conference on Engineering, Computer and Information Sciences (SIBIRCON), 2017. pp. 109-112
Review
For citations:
Tolebi G., Kurmankhojayev D. TRAFFIC DEMAND ESTIMATION BASED ON OFFLINE TRAINED ARTIFICIAL NEURAL NETWORK. Herald of the Kazakh-British Technical University. 2019;16(2):170-174.
Views:
490
Email this article 