About this Abstract |
Meeting |
MS&T23: Materials Science & Technology
|
Symposium
|
Materials Processing and Fundamental Understanding Based on Machine Learning and Data Informatics
|
Presentation Title |
B-1: Multi-objective Optimization for Improving Mechanical Properties of Aluminum Alloys: A Data Analytics Approach with Machine Learning and Genetic Algorithms |
Author(s) |
Su Jeong Kim, Yoon-Suk Choi, Su-Hyeon Kim |
On-Site Speaker (Planned) |
Su Jeong Kim |
Abstract Scope |
There is an increasing demand for high-strength lightweight materials with sufficient ductility in the transportation and aerospace industries. In the present study, a data analysis approach using machine learning (ML) and genetic algorithm (GA) was developed to search for aluminum alloys with high strength and ductility. For this purpose, the composition, heat treatment and mechanical properties of aluminum alloys were collected through the literature survey, and a data training approach was established to predict the mechanical properties. An optimum ML algorithm was chosen after trying and comparing predictabilities from a variety of ML algorithms including feature engineering and used for the multi-objective optimization based on GA. Groups of Al alloys and the heat treatment conditions expected to have high strength and ductility were generated from the converged Pareto front, and further filtered by the domain knowledge. The results were evaluated, and their physical meaning was discussed. |