Improved traffic flow estimation based on integrated learning methods

Zhi Ming Sun; Ge Ren; Yu Xin Hu; Hong Lin

doi:10.1117/12.2689892

7 September 2023 Improved traffic flow estimation based on integrated learning methods

Zhi Ming Sun, Ge Ren, Yu Xin Hu, Hong Lin

Proceedings Volume 12790, Eighth International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023); 1279020 (2023) https://doi.org/10.1117/12.2689892
Event: 8th International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023), 2023, Hangzhou, China

Abstract

This paper aims to address the current lack of accurate hourly traffic flow measurement by proposing a speed-flow measurement method based on the two integrated learning methods of LightGBM and XGBoost. The road network information, traffic speed and flow data from intelligent monitoring and recording system were integrated to build the traffic speed-flow model. The tree parzen estimator (TPE), a Bayesian optimization algorithm, was performed for the hyperparameter search of LightGBM and XGBoost, which combined with cross-validation to finally determine the optimal parameters. In addition, we compared the simulated results of LightGBM and XGBoost with BP neural network, random forest, and Van Aerde model. These results indicated that the TPE-LightGBM and TPE-XGBoost based on Bayesian optimization had more accurate and better prediction results than the other models. The MAPE of TPE-XGBoost and TPE-LightGBM decreased 12.19% and 12.26% than Van Aerde model on average, respectively. Compared with the traditional deep learning algorithm, such as BP neural network, the MAPE of TPE-XGBoost and TPE-LightGBM decreased by 4.45% and 4.52%, respectively. Our results further showed that the TPE-XGBoost had higher accuracy for estimating the flow of expressway, while the TPE-LightGBM was more suitable for analysis in the main roads and secondary roads.

Citation Download Citation

Zhi Ming Sun, Ge Ren, Yu Xin Hu, and Hong Lin "Improved traffic flow estimation based on integrated learning methods", Proc. SPIE 12790, Eighth International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023), 1279020 (7 September 2023); https://doi.org/10.1117/12.2689892

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available