Study on Methods of Traffic Estimation under Connected and Autonomous Vehicles and Manual Vehicles Mixed Environment

The real-time traffic estimation of freeways is a basic component in a traffic road platform, which is one side of an Intelligent and connected system. This paper establishes the specific traffic flow that is combined with connected and autonomous vehicles (CAVs). With the exception of the ordinary data from traffic fixed detectors, the autonomous and connected vehicles can generate data sources. In this environment, the paper proposes two models, one is based on Kalman filtering, and the other is based on BP neural network that are used to estimate road density. The Kalman filtering method is based on the data from fixed detectors, and the BP neural network method’s data sources are fixed detector and CAVs. A Vissim simulation, which is calibrated by NG-Sim data, is built to evaluate the effectiveness of the two models. Evaluation results show that the Kalman filtering model is more suitable for the situation being that the distance between fixed detectors is short, and the penetration rate of connected vehicles is below 40%. The BP neural network is appropriate for road sections, which has a high connected vehicle penetration rate (>40%), and is far away from the fixed detector. By the way, the outcome of 40% is the critical penetration rate of CAVs, which the spaces mean speed of CAVs can represent of the whole traffic flow, is also found.

Language

  • English

Media Info

  • Media Type: Web
  • Features: References;
  • Pagination: pp 185-194
  • Monograph Title: CICTP 2018: Intelligence, Connectivity, and Mobility

Subject/Index Terms

Filing Info

  • Accession Number: 01870402
  • Record Type: Publication
  • ISBN: 9780784481523
  • Files: TRIS, ASCE
  • Created Date: Jan 23 2023 12:22PM