A single-feature incident-detection algorithm based on Bayesian considerations is developed. The algorithm uses the ratio of the difference between the upstream and the downstream minute occupancies and the upstream occupancy as the traffic-flow feature followed and uses historical incident information. The historical incident data used are representative of the inner lane of the outbound Kennedy Expressway between the Chicago Loop and the Edens Expressway junction during the afternoon rush period. Mathematical expressions are developed for the distributions of the ratio from incident and incident-free data. The probabilities of incidents occurring on the outbound Kennedy are developed from available incident data. The optimal threshold to be used in the incident-detection process is determined mathematically by using Bayesian concepts. The efficiency of the algorithm is evaluated in terms of its detection rate, false-alarm rate, and mean time to detect and is compared with those of the California algorithm and two algorithms developed by Technology Services Corporation. The Bayesian algorithm compares favorably with the others with regard to detection rate (100 percent) and false-alarm rate 0.0 percent). However, its mean time to detect is greater than that of the other algorithms by almost 2.5 min. A preliminary on-line evaluation comparing the Bayesian algorithm and one of the others showed no significant differences in detection rate, false-alarm rate, and mean time to detect. /Authors/

Media Info

  • Media Type: Print
  • Features: Figures; References; Tables;
  • Pagination: pp 52-58
  • Monograph Title: Urban system operation and freeways
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00196599
  • Record Type: Publication
  • ISBN: 0309028272
  • Files: TRIS, TRB
  • Created Date: Sep 15 1979 12:00AM