Knowledge-Based Simulation of a Glass Annealing Process: An AI Application in the Glass Industry

Engineering

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 5

Applications

This paper describes a knowledge-based simulation system for a glass annealing process. The long ovens, known as lehrs, in which annealing takes place are not well understood by their operators. In fact, only a few experts can predict the effects of a change in the lehr controls. Attempts to simulate the behavior of the lehr using conventional methods have not been successful due to the size and complexity of the lehr. Our knowledge-based approach is capable of both simulating the glass temperature curve in an annealing lehr and planning the necessary lehr control settings to achieve a desired curve. It consists of two cooperating expert systems, one rule-based and the other frame-based. The system also includes a high-bandwidth graphics display which allows operators to interactively test control-setting changes and ask for the control settings which meet desired specifications. A description of the domain, a history of the development, and details of the design are all presented, along with lessons learned from the experience.