A Hierarchical Artificial Intelligence Maintenance Advisor
Abstract
Research was conducted to develop an artificial intelligence software package that assists technicians in maintenance of the Coded Pulse processor, a subsystem of the AN/SQS-53B Active Sonar System. The AI software resides on a centrally located computer and is accessible via a terminal. An architecture was demonstrated consisting of a two layer hierarchy: Functional and Isolation. The circuit topology of the Coded Pulse processor provided the basis for knowledge acquisition. The hierarchy significantly reduces the processing time required for the large knowledge base. The hierarchy replicates the system's fault isolation strategy. System monitoring information is entered by a technician into the Functional layer of the hierarchy. The Functional layer determines if the Coded Pulse processor is performing improperly and identifies the functional abnormality. Control is then directed to the Isolation layer. The Isolation layer directs the technician to perform the most efficient troubleshooting tests. Replacement recommendations are made based on adjustments made to the relative probability information during testing. Multiple faults can be identified and system testing can be redirected to other functional areas. Functional layer knowledge must be tailored to the system. Knowledge bases are required for each functional area of the Isolation layer. System architecture and results are discussed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 08, 1991
- Accession Number
- ADA238898
Entities
People
- G. J. Moss Jr
- J. A. Molnar
Organizations
- United States Naval Research Laboratory