Fault classification by a neurofuzzy network

摘要

An innovative neurofuzzy network is proposed herein for patten classification applications. A fuzzy set interpretation is incorporated into the network design to handle imprecise information. The network is equipped iwth the learning capability of neural networks to automatically deduce fuzzy membership functions and fuzzy if-then rules based on a hybrid supervised learning scheme. The network, using Gaussian neurons as membership functions both in antecedent and conseuqnetn parts, is a one-pass, on-line, and incremental learning algorithm.