A Diagnostic System Blending Deep and Shallow Reason
Abstract
Repair of the Dual Miniature Inertial Navigation System (DMINS), a navigation system used on fast attack submarines, is the responsibility of the Aerospace Guidance and Metrology Center (AGMC) located at Newark AFB, OH. Currently, diagnostics of the system are conducted by automatic test equipment (ATE). Recent plans for upgrades of the computer that drives the ATE have made possible the integration of an expert system with the ATE, thereby increasing system reliability, decreasing test times, and improving retention of site knowledge. The traditional approach to developing an expert system is to use shallow reasoning. Shallow reasoning, which encodes knowledge into a set of IF- THEN rules, is useful for incorporating diagnostic experience accumulated with a unit under test (UUT). Recent research in expert systems has emphasized the use of deep reasoning for diagnostics. A deep reasoning system diagnoses faults by constructing a model of the UUT, relying on the principle of locality and causal reasoning. Deep reasoning is suitable for a new system; if applied to an older system however, it fails to take advantage of the lessons learned by repair technicians. The focus of the thesis is the development of BDS (Blended Diagnostic System), a diagnostic system which emulates a human technician by combining shallow and deep reasoning.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1988
- Accession Number
- ADA202547
Entities
People
- James M. Skinner
Organizations
- Air Force Institute of Technology