ESTIMATION OF NETWORK-WIDE TRAFFIC SPEEDS AND FLOWS WITH HETEROGENEOUS DATA

Shaojie Liu, Mei Lam Tam, William H.K. Lam, Wei Ma, H. W. Ho, Hao Fu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

This paper proposes a model-based data-driven approach for simultaneously estimating network-wide traffic speeds and flows. In the proposed model, heterogeneous traffic data from multiple types of traffic detectors (point detectors and automatic vehicle identification detectors) on observed links are used. Fundamental diagrams of link speeds and flows are also considered. The spatial variance and covariance of link speeds and flows are incorporated into the proposed model to estimate the traffic speeds and flows on unobserved links (i.e., the links without traffic detectors) within the entire study network. The proposed model is formulated as a Kullback- Leibler (KL) divergence-based optimization problem that minimizes the KL divergence between the probability density functions of the observed and estimated link densities. A numerical example is presented to demonstrate the applications of the proposed model.

Original languageEnglish
Title of host publicationProceedings of the 26th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2022
EditorsSisi Jian, Sen Li, Hong K. Lo
Pages224-230
Number of pages7
ISBN (Electronic)9789881581402
Publication statusPublished - 2022
Externally publishedYes
Event26th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2022 - Hong Kong, Hong Kong
Duration: 12 Dec 202213 Dec 2022

Publication series

NameProceedings of the 26th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2022

Conference

Conference26th International Conference of Hong Kong Society for Transportation Studies, HKSTS 2022
Country/TerritoryHong Kong
CityHong Kong
Period12/12/2213/12/22

Keywords

  • KL divergence
  • Network-wide traffic speed-flow estimation
  • fundamental diagrams
  • spatial variance and covariance

Fingerprint

Dive into the research topics of 'ESTIMATION OF NETWORK-WIDE TRAFFIC SPEEDS AND FLOWS WITH HETEROGENEOUS DATA'. Together they form a unique fingerprint.

Cite this