Artificial Intelligence in Maintenance: Synthesis of Technical Issues.
Abstract
The principal subdisciplines of AI (e.g., expert systems, problem solving, planning, and natural language understanding) are presented as well as the larger systems engineering issues. In a chapter devoted to automated systems for managing hardware failures, the components of the failure cycle (detection, diagnosis, and repair) are described in tandem with machine approaches and applicable AI methodology. In this report, effective improvement in military maintenance is viewed to be dependent not only on automated systems but also on the development of human resources and the organizational context of maintenance. Evidence and information are provided to support the recommendation that it is possible to build more effective and less costly automated diagnostic systems only if these systems exploit human problem-solving capabilities. Four hypothetical examples of advanced systems and a comparison of human vs. machine strengths and weaknesses as problem solvers are outlined. Five research and development recommendations for the use of AI in maintenance conclude that (1) there is a good match between the need for improved maintenance and the emerging science of AI, (2) AI research should be guided by a policy of integrated diagnostics, and (3) field evaluations of AI applications should focus on organizational impact as well as technical issues, (4) programs should be targeted at both fielded systems and systems under development, (5) basic research should investigate cooperative human-machine device diagnosis problem solving and the coordination of the specification- and symptom-based approaches.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1985
- Accession Number
- ADA160863
Entities
People
- J. J. Richardson
- K. A. Dejong
- P. G. Polson
- R. A. Keller
- R. A. Maxion
Organizations
- Denver Research Institute