A scheme for knowledge representation, verification and reasoning in real time asynchronous production systems

Abstract

Expert systems for real time process monitoring need to respond to asynchronous events and reason about temporal properties of the process. This requires the ability to perform data input, event handling and temporal asynchronous reasoning during the inference cycle. Conventional expert systems do not possess these abilities. We describe a knowledge representation formalism which captures dynamic properties like external events, timing constraints and data trends in addition to diagnostic knowledge. This representation lends itself to verification of the knowledge base. We further show the implementation of a multiple rule firing asynchronous inference engine using this model.<>