An Adaptive Learning control using Self-Organizing

摘要

This paper presents an adaptive neuro-fuzzy intelligent scheme. The proposed adaptive learning control consists of the self-organizing neuro-fuzzy OSONF) ntwork and the Radial Basis Function (rBF) networks. We provide the initial SONF structure and heuristic fuzzy control rules, then apply the unsupervised learnign algorithm to structure learning if necessary. And we use the backpropagation as the adjustment of the nodes and links. So the combinations of these two algorithms can partition the input/output space in the a flexible way based on the distribution of the training data. Here the RBF network is used for system identification and provided the SNF network with the teaching signal. The proposed scheme has two improtant characteristics of adaptation and learning. The performance and applicability of the proposed scheme on injection system of furnace are presented by computer simulations.