Artificial neural networks are forms of artificial intelligence which learn correlative patterns between input and output information without a specific model. They then use the learned relationships to make predictions. An artificial neural network was constructed to recognize certain relationships in potentiodynamic polarization scans to predict if crevice corrosion, pitting, and general corrosion are possible concerns. The network so constructed was shown to be able to make appropriate predictions using scans not included in the original training. The resulting network was incorporated within an expert system to provide an easy way to input data, simple consistency checks, and to interpret the numerical output of the neural network to make the final prediction.

Subject
Noble potential, Artificial neural networks, Corrosion rate, Repassivation potentials, Crevice corrosion, Pitting, Training, Polarization, Alloys, Protection potential, Corrosion potential, Electrode potential, Types of corrosion
Keywords:
Artificial neural networks, expert system, corrosion, electrochemistry, polarization scans, localized corrosion, general corrosion, corrosion prediction, back propagation
© 1992 Association for Materials Protection and Performance (AMPP). All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means (electronic, mechanical, photocopying, recording, or otherwise) without the prior written permission of AMPP. Positions and opinions advanced in this work are those of the author(s) and not necessarily those of AMPP. Responsibility for the content of the work lies solely with the author(s).
1992
Association for Materials Protection and Performance (AMPP)
You do not currently have access to this content.