A generic framework for online knowledge-based diagnostic systems in batch chemical processes

A generic framework for a knowledge-based operator advisory system which defines the knowledge, knowledge organization and problem solving strategy for discrete-with-sufficient-time operations is presented. The operator advisory system architecture is based on the procedural decomposition achieved through a "phase-step model" of discrete processes. Operator advisories are generated for plant monitoring, fault detection and diagnosis, and corrective action planning. The architecture is implemented as a general-purpose shell and successfully demonstrated for the case of the startup of a delayed coking unit.<
>