Indirect Adaptive Control Using Fully Tuned RBFN

Abstract

Due to the adaptive characteristics of the learning process, the application of neural networks to nonlinear system identification and control has been developed in a natural way. To cope with the indirect adaptive control problem, two basic approaches have been addressed in the literature. In the first approach, some design problems are learned off-line, measuring the input-output signals and observing the plant behavior in some key situations. The control can then be implemented based on the knowledge acquired: this approach is known as an off-line training/on-line control scheme. In the second approach, an adaptive learning is implemented and the control input is determined on-line as the output of a neural network, which is called on-line learning/on-line control strategy.