Framework for a General Purpose, Intelligent
Control System for Particle Accelerators

R. T. Westervelt and W. B. Klein
Vista Control Systems Inc.

Modern control methods have proven incapable of solving very complex control problems in dynamic environments except in isolated cases. In the past, many of these complex control problems have been solved by human operators who were able to fill in incomplete information through experiential knowledge. We have developed a control system capable of tuning and controlling an accelerator beamline without human intervention. This system uses an expert system to diagnose control problems and develop high-level solution strategies. Integrated with the expert system are learning methods and new advanced technologies such as neural networks, genetic algorithms, and fuzzy logic. In this paper we describe the development of the expert system, aspects of the advanced control strategies, and demonstrate advantages of this hybrid approach over modern control methods.