KNOWLEDGE ACQUISITION AND ENGINEERING FOR STEEL BRIDGE FABRICATION

The Kansas Department of Transportation (KDOT) developed a knowledge-based system for resolving fabrication errors in steel highway bridges. The system--Bridge Fabrication Error Solution Expert System (BFX)--focuses on errors occurring with relative frequency, but not resolved by standard actions prescribed by KDOT specifications or guidelines. The authors of this paper collected example cases from fabrication shops, state inspectors' field notes, and bridge-project documents. Panel meetings and interviews were conducted with multiple-domain experts. These cases became the foundation for constructing the knowledge representation space and testing of the system. Although an ID3-based learning tool for knowledge acquisition was attempted, it was unsuccessful, leaving knowledge acquisition based primarily on explicit domain extraction. Findings revealed that BFX offered proper repair solutions two-thirds of the time and correctly reported that no solution could be properly generated using the existing knowledge base for the remaining one-third of test cases.

• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/08873801
• Corporate Authors:

  American Society of Civil Engineers

  345 East 47th Street
  New York, NY  United States  10017-2398
• Authors:
  • Melhem, H G
  • Roddis, WMK
  • Nagaraja, S
  • Hess, M R
• Publication Date: 1996-7

Language

• English

Media Info

• Features: Appendices; Figures; References; Tables;
• Pagination: p. 248-256
• Serial:
  • Journal of Computing in Civil Engineering
  • Volume: 10
  • Issue Number: 3
  • Publisher: American Society of Civil Engineers
  • ISSN: 0887-3801

Subject/Index Terms

• TRT Terms: Error analysis; Errors; Expert systems; Fabrication; Metal bridges; Solutions (Chemistry); Steel bridges
• Old TRIS Terms: Solutions
• Subject Areas: Bridges and other structures; Design; Highways; I24: Design of Bridges and Retaining Walls;

Filing Info

• Accession Number: 00728761
• Record Type: Publication
• Files: TRIS
• Created Date: Nov 28 1996 12:00AM