Circuit pack troubleshooting via semantic control. I. Goal selection

Design and analysis issues of automated diagnostic and repair stations for circuit packs are discussed. It is shown that an adoption of a control-theoretic point of view of the entire diagnostic process yields several advantages over the traditional approaches. First, the functionality of the system may be naturally modeled. Second, using the constructed model it is possible to define an adaptive control system which performs in real-time the optimal diagnostic/repair sequence of actions. Third, a natural decomposition of the control system into three cooperating yet independent subsystems is obtained, the techniques of artificial intelligence are used for efficient implementation of the subsystems. In particular, an expert systems approach is used for two of the above systems. The third (supervising) system is implemented as a heuristic search procedure. The paper concentrates on the design and implementation issues of the supervising subsystem.<>