High-Entropy alloys -- research and testing
A new approach method has been studied for the efficient and accurate prediction of high-entropy alloys (HEAs) properties. The artificial neural network (ANN) algorithm was employed to predict the mechanical properties such as yield strength, microstructure, and elongation of the alloy by training from the mole fraction and post-process information that has an influence on the mechanical properties. The mean error rate of prediction for the yield strength was 19.6%. Microstructure predictions were consistent for all test data. On the other hand, the ANN model trained only with mole fraction data had a yield strength prediction error of 33.9%. Omission of post-process data caused a decrease in the accuracy. In addition, the prediction was performed with the lasso regression model in the same way. The mean error rate of the lasso model trained with only a mole fraction was 26.1%. The lasso model trained with a mole fraction and post-process data had a yield strength prediction error of 31.1%. The linear regression equation showed limitations, as the accuracy decreased as the number of independent variables increased. As there are more variables affecting metal properties, the ANN approach is more advantageous, and the more data there are, the more accuracy increases, making it possible to design HEAs alloys that are simpler and more efficient than conventional methods. This approach predicted HEAs properties using only mole fraction and post-processing information, without the need to use conventional physicochemical theories or perform derived complex calculations.
Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).
Locate the Document
Choi, S., Yi, S., Kim, J., Shin, B., & Hyun, S. (2021). High-Entropy Alloys Properties Prediction Model by Using Artificial Neural Network Algorithm. Metals, 11(10), 1559. https://doi.org/10.3390/met11101559