An adaptive neuro-fuzzy signal processing method by using structural learning with forgetting

R. Kozma, M. Sakuma, S. Sato, M. Kitamura

Research output: Contribution to journalArticlepeer-review

2 Citations (Scopus)

Abstract

An automatic system-state monitoring method is introduced which is based on a neuro-fuzzy signal processing algorithm. The applied method utilizes massive parallel computing. Artificial neural networks act as independent neuro-agents and have the following functions in the proposed method:()) pre-processing of data in the input interface of the neuro-fuzzy system; and, (2) supporting the fuzzy rule base by making use of the information accumulated in the structure of the neural network as the result of the applied learning algorithm with forgetting. Unexperienced events are identified by the algorithm, based on the additional category ‘unknown’, which has its own membership function. The learned new feature can be used to adaptively update the knowledge-base of the monitoring system. The results are illustrated with the example of early identification of anomalies in a nuclear reactor.

Original languageEnglish
Pages (from-to)389-404
Number of pages16
JournalIntelligent Automation and Soft Computing
Volume1
Issue number4
DOIs
Publication statusPublished - 1995

Keywords

  • Anomaly detection
  • Decision making
  • Neuro-fuzzy algorithm
  • Structural learning with forgetting
  • Unexperienced event

ASJC Scopus subject areas

  • Software
  • Theoretical Computer Science
  • Computational Theory and Mathematics
  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'An adaptive neuro-fuzzy signal processing method by using structural learning with forgetting'. Together they form a unique fingerprint.

Cite this