Qualitative state-based, self-directed control systems are being introduced into the materials processing industry. These systems require expert knowledge to define the process states and appropriate control actions. Development of those states is then usually an iterative process.; In order to improve the process by which such control systems are developed, an existing discovery tool, The Scholar's Companion (TSC), was used to develop a system for the improvement of self-directed control systems. TSC, a design and data base discovery tool, was enhanced with the addition of capabilities to handle process data and control and to learn states and appropriate control responses to those states by interaction with a simulated polymer composite curing process.; The enhanced system, called the Process Experiment Theory Evaluator (PETE) was then tested using different starting models, called envisionments, and the learning and control performance evaluated. The PETE demonstrated an ability to learn, both during and between processes, and to develop a qualitative model of the process which could be used to control the curing of a polymeric composite.
展开▼
