Material Fracture Life Prediction Using Linear Regression Techniques Under High Temperature Creep Conditions

1Citations
Citations of this article
4Readers
Mendeley users who have this article in their library.

This article is free to access.

Abstract

9–12% Cr martensitic steels are widely used for critical components of new, high-efficiency, ultra-supercritical power plants because of their high creep and oxidation resistances. Due to the time consuming effort of obtaining creep properties for new alloys under high temperature creep conditions, in both short-term and long-term testing, it is often dealt with simplified models to assess and predict the future behavior of some materials. In this work, the total time to produce the material fracture is predicted according to models obtained using several linear techniques, since this property is really relevant in power plants elements. These models are obtained based on 344 creep tests performed on modified P92 steels. A multivariate analysis and a feature selection were applied to analyze the influence of each feature in the problem, to reduce the number of features simplifying the model and to improve the accuracy of the model. Later, a training-testing validation methodology was performed to obtain more useful results based on a better generalization to cover every scenario of the problem. Following this method, linear regression algorithms, simple and generalized, with and without enhanced by gradient boosting techniques, were applied to build several linear models, achieving low errors of approximately 6.75%. And finally, among them the most accurate model was selected, in this case the one based on the generalized linear regression technique.

Cite

CITATION STYLE

APA

Fernandez Martinez, R., Jimbert, P., Barbero, J. I., Callejo, L. M., & Somocueto, I. (2019). Material Fracture Life Prediction Using Linear Regression Techniques Under High Temperature Creep Conditions. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11896 LNCS, pp. 535–544). Springer. https://doi.org/10.1007/978-3-030-33904-3_50

Register to see more suggestions

Mendeley helps you to discover research relevant for your work.

Already have an account?

Save time finding and organizing research with Mendeley

Sign up for free