Conditioning of high voltage radio frequency cavities by using fuzzy logic in connection with rule based programming

Many processes are controlled by experts using some kind of mental model to decide actions and make conclusions. This model, based on heuristic knowledge, can often be conveniently represented in rules and has not to be particularly accurate. This is the case for the problem of conditioning high voltage radio-frequency cavities: the expert has to decide, by observing some criteria, if he can increase or if he has to decrease the voltage and by how much. A program has been implemented which can be applied to a class of similar problems. The kernel of the program is a small rule base, which is independent of the kind of cavity. To model a specific cavity, we use fuzzy logic which is implemented as a separate routine called by the rule base. We use fuzzy logic to translate from numeric to symbolic information. The example we chose for applying this kind of technique can be implemented by sequential programming. The two versions exist for comparison. However, we believe that this kind of programming can be powerful for controlling more complex processes, for which no mathematical formulation of parameter relations exists. The possibility exists to "tune" the controller for better performance by adapting the membership functions. An expert can do this via a simple interface.