September 9, 2024Transportation Science

Physics-Informed Machine Learning for Calibrating Macroscopic Traffic Flow Models

View Full Paper

Ask AI

Bookmark

Authors

YTYu TangQinghai University LJLi JinSupélec KÖKaan ÖzbayNew York University

Discussion

Member takes

Implication

Case study demonstrates robust traffic flow model calibration on Interstate 210 Eastbound, highlighting the resilience of physics-informed autoencoders against corrupted data.

Key Points

Physics-informed machine learning reliably calibrates macroscopic traffic flow models, achieving superior accuracy over conventional optimization during data corruption.
Case study evaluation on Interstate 210 Eastbound in California demonstrates that a deep autoencoder successfully estimates traffic parameters from flow and speed measurements.
Integrating a denoising autoencoder into the physical architecture supports robust traffic state estimation and control, even when sensor records contain missing values.

Cite This Study

Tang et al. (2024) studied this question.

synapsesocial.com/papers/68e58feab6db64358752b4bd https://doi.org/10.1287/trsc.2024.0526

View Full Paper

Ask AI

Bookmark

Implication

Case study demonstrates robust traffic flow model calibration on Interstate 210 Eastbound, highlighting the resilience of physics-informed autoencoders against corrupted data.

Physics-Informed Machine Learning for Calibrating Macroscopic Traffic Flow Models

Discussion

Member takes

Implication

Key Points

Cite This Study

Physics-Informed Machine Learning for Calibrating Macroscopic Traffic Flow Models

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider