Multi-label phase-prediction in high-entropy-alloys using Artificial-Neural-Network

摘要

A novel Artificial-Neural-Network architecture has been developed, for the first time, which predicts eight coexisting-phases present in high-entropy-alloys. The model considers composition and processing-route to compute physical and thermodynamic-parameters of the alloy. The Artificial-Neural-Network uses these parameters and predicts the phases in a given alloy. Validation reveals prediction accuracy of 87.083%. This modeling technique can find potential application in designing new high-entropy-alloys and choosing their processing-routes. (C) 2020 Elsevier B.V. All rights reserved.