Explainable machine learning and intelligent optimisation algorithms synergistically driven prediction of fracture strength in amorphous alloys

Xiaowei Liu, Yulong Huang, Zhaosheng Xu, Jianbang Liu,
Explainable machine learning and intelligent optimisation algorithms synergistically driven prediction of fracture strength in amorphous alloys,
Journal of Alloys and Compounds,
Volume 1043,
2025,
184263,
ISSN 0925-8388,
https://doi.org/10.1016/j.jallcom.2025.184263.
(https://www.sciencedirect.com/science/article/pii/S0925838825058256)
Abstract: This paper establishes a collaborative research framework for predicting the fracture strength of amorphous alloys, integrating intelligent optimization algorithms, interpretable machine learning, and experimental validation. Specifically, the iterative update mechanism of the whale optimization algorithm is enhanced by incorporating a nonlinear convergence factor, dynamic inertia weighting, and a Lévy flight strategy. These modifications significantly improve the dynamic balance between the algorithm's global exploration and local exploitation capabilities, thus enabling efficient hyperparameter optimization for the machine learning model. Furthermore, a quantitative feature contribution assessment system is developed based on the Shapley Additive Explanations method. This system elucidates the influence of input variables on fracture strength by evaluating both global feature importance and local sample sensitivity. Finally, based on three experimentally prepared amorphous alloy samples and 12 published data collected in the literature, a fracture strength dataset covering multi-component systems is established to verify the generalisation ability of the constructed model in strength prediction.
Keywords: Machine learning; Amorphous alloy; Fracture strength; Whale optimisation algorithm; Shapley additive explanations