The use of neural networks to recognise and predict traffic congestion

M. S. Dougherty, H. R. Kirby, R. D. Boyle

Research output: Contribution to journalArticlepeer-review

Abstract

This paper shows how neuro computing can assist the pattern recognition of two main road system problems, namely the state of a road system and short term forecasting. Recognising congestion is subjective and is difficult to express such decision making algorithmically. There is also a need for rapid short term prediction. The benefit of a neural network is that it absorbs patterns in data and so can learn to generalise. The main features of a neural network approach, the trials of its application to a congestion recognition problem (based on Leicester SCOOT data), and the trials of short term forecasting of flows (again with SCOOT data) are all described. Included are graphs of predictions made. The paper goes on to discuss whether neural networks can be used to infer parameters not directly measureable on the street. -D.Jarratt

Original languageEnglish
Pages (from-to)311-314
Number of pages4
JournalTraffic Engineering & Control
Volume34
Issue number6
Publication statusPublished - 1993
Externally publishedYes

Keywords

  • artificial intelligence
  • computer programs
  • forecasting
  • information processing
  • traffic congestion

Fingerprint

Dive into the research topics of 'The use of neural networks to recognise and predict traffic congestion'. Together they form a unique fingerprint.

Cite this