This report describes the results of a project involving the development of an expert diagnostic system for marine diesel engines. The purpose of the project was to develop a practical expert system aboard a modern, automated diesel-powered vessel to assist maintenance personnel in diagnosing existing engine failures and predicting future faults. Research activities involved developing robust diagnostic and automated reasoning techniques based on recent advancements in the field of expert systems. Alternative diagnostic reasoning techniques, such as rule-based inferencing, neural networks, structural and behavioral models, and probabilistic techniques, were investigated before settling on a final design involving probabilistic neural networks. Prototype systems were placed into operation aboard four large containerships powered by two-stroke, slow-speed diesel engines. The system configuration included data acquisition interfaces to existing engine monitoring and combustion analysis systems, as well as operation on a shipboard local area network.

  • Corporate Authors:

    American President Lines

    1111 Broadway
    Oakland, CA  United States  94607

    MACSEA Limited

    163 Water Street
    Stonington, CT  United States  06378

    Maritime Administration

    1200 New Jersey Avenue, SE
    Washington, DC  United States  20590
  • Authors:
    • Logan, K P
  • Publication Date: 1992-9


  • English

Media Info

  • Pagination: 82 p.

Subject/Index Terms

Filing Info

  • Accession Number: 00672416
  • Record Type: Publication
  • Report/Paper Numbers: MA-RD-840-92002
  • Contract Numbers: MA-12096
  • Files: TRIS, USDOT
  • Created Date: Jan 11 2003 12:00AM