AN APPLICATION OF ARTIFICIAL NEURAL NETWORKS TO PISTON RING FRICTION LOSSES PREDICTION
Mots-clés :
piston ring, friction losses, artificial neural networkRésumé
A significant share of the total power loss in a modern automotive engine is due to the friction interaction between the piston ring pack and the cylinder wall. This paper presents the results of the simulations on the friction interaction top ring/cylinder wall in a SI engine taking into account the mixed lubrication (ML) regime and considering different engine operating conditions, lubricant viscosity, surface roughness. An Artificial Neural Network model and a procedure are presented in order to predict and optimise the friction losses. The ANN model allows for identifying the factors with the biggest influence on the friction loss evolution, while it can be used for finding the combinations of inputs factors leading to the lowest average friction coefficient value.
Téléchargements
Publiée
Numéro
Rubrique
Licence
All rights are reserved for this publication, which is copyright according to the Romanian law of copyright. Excepting only any fair dealing for the purpose of the private study, research, review, comment and criticism, no part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, mechanical, electrical, electronic, optical, photocopying, recording or otherwise, without the prior express permission of the copyright owners.
