Short Term Prediction Services From Pro-Active Urban Traffic Management

The usage of urban and interurban roads is still expanding in most European countries. This leads to road networks running out of capacity, increasing congestion rates and a setback in urban accessibility. Road operators only have limited sources available such as loop detectors, traffic controllers and (in some cases) speeds derived from floating car data to provide an accurate and consistent picture of the actual state of the road network. As a result, traffic management measures such as traffic controllers, route guidance and corridor management are not used to their full potential. To unleash its full potential, more accurate information of the actual traffic state is required but, above all, it is required to know about near future traffic situations to be able to act in advance. In cooperation with Technolution and the municipality of Deventer, with approximately 100.000 inhabitants, a short term prediction algorithm has been developed and integrated within the Network Management System of the city. This allows the city’s network operators to anticipate pro-active on actual and near future traffic situations. Operators have live access to traffic state predictions of all main roads within the city including its connections to the highway network and are able to inform drivers about actual travel times to and from the city center using Dynamic Route Information Panel Signs (DRIPS). The authors paper and presentation will describe the approach used and the results regarding the quality of the predictions as well as the lessons learned from a network operator perspective. The authors model based short term prediction algorithm has been expanded with urban functionalities and has been deployed, assessed and actually being used in the operational context of urban traffic management. Both for demand and for supply estimation, urban environments result in more demanding algorithms than highway environments. Urban environments are often less well measured with traditional loop detectors and floating car data for urban environments is of less quality due to lower flow rates. Moreover, aspects regarding route choice and a realistic ratio between origins and destination have more impact within an urban setting. On the other hand, urban environments offer new types of data sources (i.e. traffic signal and parking data) to include within data fusing processes and to improve accurate demand estimation. Regarding supply estimation, capacities of intersections are the prime sources that affect traffic flow quality in urban environments, in contrary to the capacity of merging and diverging sections which primarily affect traffic flow quality in highway settings. The authors have enhanced the authors short term prediction algorithm for urban environments to be able to accurately handle such urban conditions. The prediction algorithm is based on four key features; data fusion, artificial intelligence (AI), traffic flow simulation and real-time estimation of fundamental diagrams as well as junction capacities. Data fusion and AI-techniques are used to aggregate and fuse data from various sources into reliable and consistent traffic state data. Historic traffic flow patterns are registered and used to enrich actual measurements for more accurate demand/OD-matrix estimation (for near future minutes). A feed-forward neural network has been included within this algorithm to provide detailed predictions for in- and outflow of parking garages. A macroscopic traffic propagation model within OmniTRANS transport planning software is used to process this traffic state data and complete a full common operational picture as well as near future predictions up to 30 minutes in advance including speeds, flows and travel times. Fundamental traffic flow theory is used to continuously (typically every minute) estimate and update actual road and junction capacities within the model environment.

  • Record URL:
  • Supplemental Notes:
    • Abstract used by permission of Association for European Transport. Alternative title: Short Term Prediction Services for Pro-Active Urban Traffic Management
  • Corporate Authors:

    Association for European Transport (AET)

    1 Vernon Mews, Vernon Street, West Kensington
    London W14 0RL,    
  • Authors:
    • Suijs, Leon
    • Mein, Edwin
  • Conference:
  • Publication Date: 2020


  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; References; Tables;
  • Pagination: 14p
  • Monograph Title: European Transport Conference 2020

Subject/Index Terms

Filing Info

  • Accession Number: 01768533
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Feb 17 2021 2:45PM