Machine learning to accelerate alloy design

• Main Authors: Liaw, Peter K, Steingrimsson, Baldur Andrew, Kim, Duckbong, Kulkarni, Anand A, Fan, Xuesong
• Format: Patent
• Language: English
• Published: 27.02.2024
• Subjects: ADDITIVE MANUFACTURING TECHNOLOGY; ADDITIVE MANUFACTURING, i.e. MANUFACTURING OFTHREE-DIMENSIONAL [3-D] OBJECTS BY ADDITIVE DEPOSITION, ADDITIVEAGGLOMERATION OR ADDITIVE LAYERING, e.g. BY 3-D PRINTING,STEREOLITHOGRAPHY OR SELECTIVE LASER SINTERING; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING; CALCULATING; CASTING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTEDFOR SPECIFIC APPLICATION FIELDS; MAKING METALLIC POWDER; MANUFACTURE OF ARTICLES FROM METALLIC POWDER; PERFORMING OPERATIONS; PHYSICS; POWDER METALLURGY; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDEDFOR; SHAPING OR JOINING OF PLASTICS; TRANSPORTING; WORKING METALLIC POWDER; WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE, IN GENERAL

This invention presents an innovative framework for the application of machine learning for identification of alloys or composites with desired properties of interest. For each output property of interest, we identify the corresponding driving (input) factors. These input factors may include the material composition, heat treatment, process, microstructure, temperature, strain rate, environment or testing mode. Our framework assumes selection of optimization technique suitable for the application at hand and data available, starting with simple linear, or quadratic, regression analysis. We present a physics-based model for predicting the ultimate tensile strength, a model that accounts for physical dependencies, and factors in the underlying physics as a priori information. In case an artificial neural network is deemed suitable, we suggest employing custom kernel functions consistent with the underlying physics, for the purpose of attaining tighter coupling, better prediction, and extracting the most out of the-usually limited-input data available.