The accelerated discovery of multi-component alloys with tailored properties poses a critical challenge in materials science, demanding both computational precision and efficiency. We present an integrated ICME framework that leverages state-of-the-art Bayesian optimisation (BO) protocols [1,2,3], grounded in active learning (AL) and extended to continuous design spaces. This enables flexible and data-efficient exploration of complex alloy systems. Benchmarking of advanced multi-objective BO acquisition functions-including TSEMO, parEGO, qEHVI, qNEHVI, and qNparEGO-demonstrates robust performance in navigating trade-offs across multiple targeted properties. The framework incorporates CALPHAD-based simulations for accurate predictions of phase stability and thermodynamic behavior, autonomously selects optimal acquisition strategies, and employs Design of Experiments (DOE) to minimise experimental iterations. A pool-based AL approach facilitates concurrent high-throughput computational and experimental workflows, while multi-scale modelling and uncertainty quantification ensure performance reliability and prediction robustness. Real-time machine learning integration enables iterative feedback from experiments, continuously refining the alloy optimisation process. This holistic approach offers a scalable, data-driven pathway to accelerate the design and qualification of advanced alloys.

References

[1]. O. Mamun, M. Bause, and B. S. M. Ebna Hai; Accelerated development of multi-component alloys in discrete design space using Bayesian multi-objective optimisation, Machine Learning: Science and Technology, 2025, volume 6, number 1, 015001, DOI: 10.1088/2632-2153/ada47d

[2]. O. Mamun, A. Mukherjee, U. Saikia, and B. S. M. Ebna Hai; AI-Enhanced ICME Framework for Materials Design, MaterialsWeek 2025, 02-04 April 2025, Frankfurt am Main, Germany.

[3]. O. Mamun, and B. S. M. Ebna Hai; AI-Enhanced Integrated Computational Materials Engineering Framework for Efficient Alloy Design in Continuous Design Spaces. In: 8th World Congress on Integrated Computational Materials Engineering, TMS Specialty Congress 2025, 15-19 June 2025, Anaheim, California, USA.