A Recurrent Neural Network Approach to Network-wide Traffic Signal Control

This paper presents a recurrent neural network approach to network-wide traffic control, referred to as NEURA-NESC. The system is based on a traffic flow model, which comprises two recurrent neural networks. The control model of NEURA-NESC extends this traffic model by two error-propagation networks that are related to the most essential parameters of traffic signal control: green time split and offset. The time values correspond directly with link capacities that are adjusted according to the minimization of some given objective function. Simulation investigations demonstrate the efficiency of the proposed approach. Furthermore, in the complete model a very close cooperation between network-wide dynamic traffic assignment and signal control is achieved.


  • English

Media Info

  • Media Type: DVD
  • Features: Figures; References; Tables;
  • Pagination: 23p
  • Monograph Title: TRB 89th Annual Meeting Compendium of Papers DVD

Subject/Index Terms

Filing Info

  • Accession Number: 01152864
  • Record Type: Publication
  • Report/Paper Numbers: 10-0399
  • Files: TRIS, TRB
  • Created Date: Jan 25 2010 10:13AM