Machine Learning and Data Analytics for Design and Manufacturing of High-Entropy Materials Exhibiting Mechanical or Fatigue Properties of Interest

Abstract

This chapter presents an innovative framework for the application of machine learning and data analytics for the identification of alloys or composites exhibiting certain desired properties of interest. The main focus is on alloys and composites with large composition spaces for structural materials. Such alloys or composites are referred to as high-entropy materials (HEMs) and are here presented primarily in context of structural applications. For each output property of interest, the corresponding driving (input) factors are identified. These input factors may include the material composition, heat treatment, manufacturing process, microstructure, temperature, strain rate, environment, or testing mode. The framework assumes the selection of an optimization technique suitable for the application at hand and the data available. Physics-based models are presented, such as for predicting the ultimate tensile strength (UTS) or fatigue resistance. We devise models capable of accounting for physics-based dependencies. We factor such dependencies into the models as a priori information. In case that an artificial neural network (ANN) is deemed suitable for the applications at hand, it is suggested to employ custom kernel functions consistent with the underlying physics, for the purpose of attaining tighter coupling, better prediction, and for extracting the most out of the -usually limited -input data available. 1 The chapter is dedicated to the loving memory of Dr. Steingrimur Baldursson, Professor Emeritus at the University of Iceland (1930Iceland ( -2020. In 1958, prior to returning back to Iceland to take on a position within the Faculty of Sciences at the University of Iceland, Dr. Baldursson completed a doctorate from the University of Chicago, under guidance of Prof. Joseph E. Mayer and Prof. Maria G. Mayer, a 1963 Nobel Laureate in Physics.

Physics‐Based Machine‐Learning Approach for Modeling the Temperature‐Dependent Yield Strength of Superalloys

“…Hence, we have suggested a framework for joint optimization. [2][3][4] For instance, there typically is a trade-off between the ductility and the strength of the superalloys. Figure 1 presents an inverse design process, where components of the design are explicitly calculated from the target performance metrics provided.…”

“…[4] accounts for the dependence between input sources, and Section 4.5 of ref. [4] addresses the expected dependence of the US on the individual input sources listed. The prediction model can be summarized as [ 3,13 ] …”

“…Note, furthermore, that for a new alloy system, T m does not have to be known experimentally in advance either; a rough estimate for T m may be obtained by using “the rule of mixing” and a more refined estimate obtained by employing the Calculation of Phase Diagram simulations. [ 4 ]…”

“…As opposed to specifically applying ML, narrowly defined in terms of multivariate regression, single-layer or multi-layer neural networks, Bayesian graphical models, random forest models, support vector machines (SVMs), or decision trees, to the identification of superalloy compositions of interest, we reformulate the task in the broader context of engineering optimization. [2][3][4] We recommend selecting an optimization technique suitable for the application at hand and the data available, but we certainly include ML in the consideration. Our rationale is based in part on observations from Agrawal et al [5] According to Table 2 and Figure 5 of ref.…”

“…For further background information on the application of ML to the modeling of mechanical properties of alloys and on the motivation for coupling with physics-based modeling approaches, refer to refs. [2][3][4].…”