Dynamic process diagnosis via integrated neural networks

Abstract A generic scheme of integrated artificial neural networks has been studied in this work for the purpose of fault detection and diagnosis in dynamic systems with varying inputs. Two general types of neural models, i.e. the feedforward networks (FFNs) and the recurrent networks, were integrated in the proposed fault monitoring system. To demonstrate the utility of the proposed methodologies, extensive experimental studies have been carried out on a pilot plant which simulates the operation of a semi-batch storage system. It can be observed that the predictions of the normal system behavior are very accurate and, also, the diagnostic conclusions obtained with the integrated networks are highly reliable.