A framework for knowledge-based computer-integrated manufacturing

A description is given of an evolving framework for applying knowledge systems in a manufacturing environment. This framework consists of a set of object-oriented tools for modeling key elements of the environment (including processes, equipment, facilities, and operational procedures) and a complementary set of application-specific shells that use the models to perform common manufacturing tasks such as monitoring, diagnosis, control, simulation, and scheduling. The testbed used is the semiconductor fabrication line at Stanford's Center for Integrated Systems. The key concept is a symbolic model of the manufacturing environment that can be created and maintained by manufacturing personnel. This model provides an open, unified framework that supports a variety of expert systems and integrates them with existing computer-integrated manufacturing (CIM) packages. The models provide a single consistent representation of the manufacturing domain that is shared by all applications. The shells embody methodologies, algorithms, and knowledge representations that are task-specific, but largely independent of the domain. Customized with domain-specific expertise, the shells will work for any environment that can be modeled.>