SBIR-STTR Award

To which conventional control methods cannot be satisfactorily applied. Advanced information technologies such as expert systems and neural networks have been applied separately to these problems with limited success. Few, if any, of these advanced information technologies have been applied for general use or in a manner usable by multiple accelerator installations. In Phase I existing neural network and expert system technologies will be coupled to provide a general purpose, intelligent control system. This approach offers a feasible solution to the complex control problems encountered in accelerator controls. These technologies will be integrated within the framework of an existing commercial, general purpose control system that is extensively used by the accelerator control community. Also, typical and realistic scenarios in accelerator control will be defined using a commercial accelerator modeling code that will be enhanced to incorporate realistic noise effects. These scenarios will be used as a test platform for the general purpose control framework being developed as part of this work.

Accelerator control systems have long applied advanced computing technologies to the demanding requirements of state-of-the-art physics research facilities. In theory, an intelligent accelerator control system should improve efficiency, maintainability, and delivery of beam on target. However, intelligent control systems have lagged behind other state-of-the-art computing technologies for a variety of reasons. Computing power has until recently been insufficient to meet accelerator control demands and the complexity of the problems encountered in controlling large accelerator facilities was typically too great for a general-purpose, intelligent control system to address fully. Recent advances in computer hardware and control software performance and capabilities have lifted these limitations and made it possible to develop advanced systems providing cost-effective, intelligent accelerator control. This project will develop a general purpose, intelligent control system for use in accelerator facilities. In Phase I, different tuning problems proved amenable to different types of solution strategies. Some techniques were shown to work well for some problems and poorly for others. An expert system could chose, from a supported set of heuristics, the method best suited to solve the current control problem. Phase II will develop, test, and distribute a complete prototype for an intelligent accelerator control system. The intelligent controller will be tested under actual operating conditions at representative accelerator facilities. Data on operations from these facilities will be used to develop and fine-tune the heuristic control techniques. The set of control scenarios, developed during Phase I, will be codified for distribution among the accelerator controls and artificial intelligence communities. The scenarios developed will include simple steering, periodic lines, matching beam transformer, general matching and the achromatic bend. We expect these scenarios to provide a common frame of reference for work on advanced accelerator control systems.Anticipated Results/Potential Commercial Applications as described by the awardee: The result of the project will be a viable prototype for an intelligent control system for particle accelerators. This prototype will serve as the basis for the development of a commercially available general purpose intelligent control system. This general purpose intelligent control system will be useful in a wide range of industrial control applications.