Fault Detection and Isolation Using Dynamic Co-Active Neuro-Fuzzy Systems

The contribution addressed by this paper refers to the development of a new dynamic co-active neuro-fuzzy system and its application to fault detection and isolation of an evaporation station. The training of the neuro-fuzzy system is done by a hybrid learning. This is based on a fuzzy clustering algorithm to determine the number of fuzzy rules and the values of the premise parameters, and steepestdescent algorithms to basically determine the consequent parameters. The developed dynamic co-active neuro-fuzzy system is then tested in the framework of an experimental case study. This refers to the sensor and actuator fault diagnosis of an evaporation station from a sugar factory. For this purpose, an extended neuro-fuzzy generalised observer scheme is designed to generate the residuals (symptoms) in the form of the one-step-ahead prediction errors. These are then processed by a neural classifier in order to take the appropriate decision regarding the actual behaviour (normal or faulty) of the process.