Robust fault detection using neuro-fuzzy networks

Authors:	Kowal Marek, University of Zielona Gora, Poland Korbicz Józef, University of Zielona Gora, Poland
Topic:	6.4 Safeprocess
Session:	Fusion of Analytical and Soft Computing Methods in Fault Diagnosis
Keywords:	fault detection, neuro-fuzzy network, uncertainty, bounding error approach

Abstract

The paper focuses on the problem of robust fault detection using neuro-fuzzy model based strategies.The main objective of the work is to show how to employ bounding error approach to determine the uncertainty of the neuro-fuzzy model and next utilize this knowledge for robust fault detection. The paper presents also how to tackle the problem of choosing the right structure of the neuro-fuzzy models. Proposed algorithms are applied to fault detection in the valve that is the part of the technical installation at the Lublin sugar factory. Experimental results presented in the final part of the paper confirms the effectiveness of the proposed methods.