Quantitative Structure-Activity Relationships (QSAR) based models have been widely used for predicting corrosion inhibition performance of metals. However, one of the major limitations in these studies is that the authors have restricted themselves to use only a single class of molecules having similar molecular structure. In this study, a computational end-to-end framework was developed to investigate the properties of organic corrosion inhibitors which are responsible for inhibition of steel in acidic solution. The framework consists of modules like data preprocessing, descriptor selection and model building. A robust predictive model for multiple class of corrosion inhibitors was developed using advanced machine learning algorithm such as gradient boosting machine (GBM), random forest, support vector machines (SVM) etc. The descriptors were selected using novel integrated ensemble technique. The model based on GBM algorithm was able to predict the corrosion inhibition efficiency of inhibitors with significantly higher accuracy.

Subject
Acids, Correlation, Machinery, Algorithms, Molecular structure, Experimental data, Corrosion modeling, Inhibition efficiency, Corrosion inhibitors, Steel, Carbon, Computer based testing, Training
Keywords:
Corrosion, inhibitor, steel, machine learning, QSAR
© 2019 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).
2019
Association for Materials Protection and Performance (AMPP)
You do not currently have access to this content.