Artificial Intelligence Technology for the Maintainer's Associate

Abstract

Futue trends in technological sophistication, personnel resources, and warfare scenarios are expecteL.to aggravate the Ja4 tef.ass situation.In view of) these problems and the current Depatment of Defense 1 olicies regarding Integrated diagnostics and a reduced reliance on paper-'based documentation, the concept of an Interactive, portable, computer-based maintainer's associate h-bei proposed.The purpose oT this effort was-tS developAthe technology for the Maintainer's Associate based on artificial intelligence techniques and demonstrate a prototype system in the field. (Concluded)In the course of this work, several + -7wtw issues were examined: hybrid diagnostics, knowledge engineering costs, user interfaces, and the integration of training and job aiding.The term whybrid 1 diagnostics. refers to the utilization of multiple sources of knowledge in the development of maintenance expert systems, in particular (a)'-dependency modeling fpatett+e+1-y.erivable from engineering databases)--Ind (b) heuristic expertise of field technicians. In the area of dependency modeling, one source of knowledge identified for the prototype was the test program set of the automatic test station. This information provided the specific measurement values and locations necessary for making measurements during troubleshooting.nowledge engineering costs were controlled through use of these test program sets and the F development of a "glass box"'editor which permitted knowledge base modifications during program operation.

Computational Approaches for Analyzing Tradeoffs Between Training and Aiding

“…Intelligent systems react to human performance and adapt the nature of the interaction accordingly. A variety of such systems are under development which integrate training and aiding within a single support system (Johnson, 1981;Keskey & Sikes, 1987;Kline & Lester, 1988;Richardson et al, 1986).…”