Browse by author
Lookup NU author(s): Professor Ehsan Mesbahi
Full text for this publication is not currently held within this repository. Alternative links are provided below where available.
This work is related to the use of artificial neural networks (ANNs) for identification of selected faults in a six-cylinder Ruston 6APC medium-speed diesel engine. Various faults were implemented on the Ruston engine and an extensive set of data at different operational conditions was acquired and analysed. The selected data with their known engine operating condition were used to train neural networks, Two different approaches in diesel engine fault diagnosis were utilised, firstly the fault pattern method, and secondly, the residual method. The trained neural networks were evaluated using a separate set of test data unseen by the neural networks during training, as well as the whole data set. The results were analysed to determine the reliability of using the above-mentioned methods and the distribution of the errors was addressed in order to identify which operational conditions were hardest to diagnose. Furthermore, one of these methods was extended, allowing human intervention in the fault identification process through a graphical interface. The final results show that faulty and healthy conditions can be identified by means of ANNs with a target ratio higher than 99% on the whole data set.
Author(s): Mesbahi E, Arriagada J, Assadi M, Ghorban H
Publication type: Article
Publication status: Published
Journal: Journal of Marine Design and Operations
Year: 2004
Issue: 6B
Pages: 29-37
Print publication date: 01/01/2004
ISSN (print): 1476-1556
ISSN (electronic):
Publisher: Institute of Marine Engineering, Science and Technology
