Intelligent Signal Processing for Active Control
P. A. Ramamoorthy · 1992
This research is concerned with the use of neural architectures and fuzzy expert systems in nonlinear system identification and in the control of such systems. In particular, on-line identification/modeling is considered. The research has resulted in a technique where the network can evolve (in size) in time-so as to provide an optimal model/controller. Also an adaptive algorithm, which is less sensitive to initial values of the weights and the learning rate, has been developed. We have also established a common framework between neural networks and fuzzy expert systems and developed a neuro-fuzzy architecture that retains the best of the two areas. The use of the architectures and the adaptation algorithm has been demonstrated on a number of applications.