Self-Organization and Basis Functions of Neural Network Controllers
Takayuki Yamada, Norifumi Yasue and Takenori Morimitsu
NTT Access Network Systems Laboratories, Shirane 162, Ooaza Shirakata, Tokaimura, Naka-gun, Ibaragi-Pref., 319-11, Japan
This paper discusses a relationship between a controller for an unknown nonlinear plant and a self-organization capability which changes the structure of the controller. The controllers using both a neural network and a basis function have the potential to realize a nonlinear control system with self-organization capability because they can change their structure through the change of the number of neurons or basis functions.