Aluminium Alloy Design and Discovery using Machine Learning

Aluminium Alloy Design and Discovery using Machine Learning J. Mangos and N. Birbilis * College of Engineering and Computer Science, The Australian National University, Acton, A.C.T., 2601, Australia. *[email protected] Abstract The traditional design and development of metallic alloys has – to date – taken a hill-climbing approach, with incremental advances. Throughout the last century, aluminium (Al) alloy design has been essentially empirical and iterative, based on lessons learned from in-service use and human experience. Incremental alloy development is costly, slow, and doesn't fully harness the data that exists in the field of Al-alloy metallurgy. In the present work, an attempt has been made to utilise a data science approach to develop a machine learning (ML) model for Al-alloy design. An objective-optimisation process has also been developed, to exploit the ML model, for user experience and practical application. A successful model was developed and presented herein, along with the open-access software. Keywords: Alloy design, machine learning, aluminium, corrosion, sensitization, mechanical properties.

Aluminium Alloy Design and Discovery using Machine Learning

Documents

alloy design

machine learning

aluminium

corrosion

sensitization

mechanical properties