NEURAL NETWORK ESTIMATION OF WATERWAY LOCK SERVICE TIMES

Good service-time estimates at locks are essential for evaluating waterway performance, planning improvements, and controlling operations. Difficulties in estimation are due to great variations in lock characteristics, vessel characteristics, operating options, and environmental conditions. In this study several artificial neural network models for lock service-time estimation are developed and compared. Results show that simple artificial neural network models yield lower prediction errors than simple regression models, that systematic removal of outliers can reduce the number of artificial neural network prediction errors, and that combined service-time models for locks with dissimilar chambers can be obtained without unreasonably compromising accuracy.

Language

  • English

Media Info

  • Features: Figures; References; Tables;
  • Pagination: p. 36-43
  • Monograph Title: Artificial intelligence and geographical information
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00714941
  • Record Type: Publication
  • ISBN: 0309061636
  • Files: TRIS, TRB
  • Created Date: Dec 20 1995 12:00AM