Neural net control vs. anticipatory fuzzy control for a flexible beam: a comparison

An anticipatory fuzzy controller has been developed which combines traditional fuzzy logic with feedback concepts and a system model to predict system behavior. Fuzzy rules in the system are `nested´ so that it is not necessary to evaluate all rules at each time step if some smaller subset of rules will produce acceptable system behavior. A neural network has been trained to emulate this controller, to eliminate the necessity of using a model in the final system. The results of the anticipatory fuzzy controller with perfect model prediction and the neural net trained to replace this controller are compared here