View in EDS

Integrated design framework for titanium aluminides through interpretable machine learning.

Saved in:
Bibliographic Details
Title: Integrated design framework for titanium aluminides through interpretable machine learning.
Authors: Padhy, Shakti P.1 (AUTHOR), Davidson, Karl P.2 (AUTHOR), Tan, Li Ping3 (AUTHOR), Varma, Vijaykumar B.4 (AUTHOR), Sharma, Vinay K.5 (AUTHOR), Tan, Xiao6 (AUTHOR), Wei, Yuefan7 (AUTHOR), Xu, Xuesong8 (AUTHOR), Hippalgaonkar, Kedar3,9 (AUTHOR), Jhon, Mark H.10 (AUTHOR), Ramanujan, R.V.1,3 (AUTHOR) ramanujan@ntu.edu.sg
Source: Journal of Alloys & Compounds. Dec2025, Vol. 1047, pN.PAG-N.PAG. 1p.
Subjects: Titanium aluminides, Machine learning, Surrogate-based optimization, Aerospace technology, Alloys, Hardness, Materials science
Abstract: Ti-Al based alloys are high temperature structural materials used in extreme aerospace applications, such as jet engine blades. However, conventional discovery of new alloy compositions with superior properties has been slow and resource-intensive. To accelerate this, a novel interpretable machine learning (ML) framework was developed to identify novel alloy compositions with promising properties. While our integrated ML framework builds upon prior work in materials design, its novelty lies in the systematic application to titanium aluminide alloys and the specific use of explainable AI (XAI) techniques, particularly Shapley Additive exPlanations (SHAP), to interpret feature influence and guide the definition of the search space for one-shot multi-property Bayesian optimization (MPBO). This framework also encompasses comprehensive data collection from literature, an ML-based imputation strategy to handle data sparsity, and robust multi-property regression algorithms. Alloy C-I (Ti- 49.4Al- 3.5Cr- 2.9Nb) and Alloy C-II (Ti- 47.2Al- 3.8Cr- 3Nb) were predicted using this framework. Validation experiments show that these compositions demonstrated superior room-temperature yield and tensile strengths in both tension and compression tests, and also superior hardness, compared to a reference Ti-4822 alloy prepared under identical laboratory conditions. Thus, our approach advances the development of high-performance titanium alloys and exemplifies the integration of ML into materials discovery. [Display omitted] • Integrated design framework for titanium aluminides developed. • Interpretable machine learning (ML) used for materials discovery. • Database of 1937 Ti-Al alloy data points curated from literature. • ML framework identifies novel promising alloy compositions. • New alloys show enhanced strength and hardness vs. Ti-4822. [ABSTRACT FROM AUTHOR]
Copyright of Journal of Alloys & Compounds is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:Ti-Al based alloys are high temperature structural materials used in extreme aerospace applications, such as jet engine blades. However, conventional discovery of new alloy compositions with superior properties has been slow and resource-intensive. To accelerate this, a novel interpretable machine learning (ML) framework was developed to identify novel alloy compositions with promising properties. While our integrated ML framework builds upon prior work in materials design, its novelty lies in the systematic application to titanium aluminide alloys and the specific use of explainable AI (XAI) techniques, particularly Shapley Additive exPlanations (SHAP), to interpret feature influence and guide the definition of the search space for one-shot multi-property Bayesian optimization (MPBO). This framework also encompasses comprehensive data collection from literature, an ML-based imputation strategy to handle data sparsity, and robust multi-property regression algorithms. Alloy C-I (Ti- 49.4Al- 3.5Cr- 2.9Nb) and Alloy C-II (Ti- 47.2Al- 3.8Cr- 3Nb) were predicted using this framework. Validation experiments show that these compositions demonstrated superior room-temperature yield and tensile strengths in both tension and compression tests, and also superior hardness, compared to a reference Ti-4822 alloy prepared under identical laboratory conditions. Thus, our approach advances the development of high-performance titanium alloys and exemplifies the integration of ML into materials discovery. [Display omitted] • Integrated design framework for titanium aluminides developed. • Interpretable machine learning (ML) used for materials discovery. • Database of 1937 Ti-Al alloy data points curated from literature. • ML framework identifies novel promising alloy compositions. • New alloys show enhanced strength and hardness vs. Ti-4822. [ABSTRACT FROM AUTHOR]
ISSN:09258388
DOI:10.1016/j.jallcom.2025.184937