Litcius/Paper detail

Prediction of mechanical properties and fractography examination of AZ91 magnesium composites reinforced with graphene using a random forest machine learning model: experimental validation

Song‐Jeng Huang, Yudhistira Adityawardhana

2025Archives of Civil and Mechanical Engineering12 citationsDOIOpen Access PDF

Abstract

Abstract Machine learning (ML), a prominent branch of artificial intelligence, is increasingly applied in material design, particularly for magnesium composites. In this study, random forest models were used to predict mechanical properties and fractographic behavior using regression classification, respectively. Both the regression and classification models of the random forest demonstrated high accuracy in predicting new optimal mechanical properties for a composite containing 0.16 wt% graphene, which was enhanced through T6 heat treatment and equal channel angular pressing (ECAP). The predictions were further validated through laboratory experiments. Although not all predicted mechanical property values exceeded the optimal values obtained from the experiments, the strain-hardening capacity of the ML-recommended samples was higher than that of the experimental samples. In addition, the predicted surface features using fractography closely matched the experimental validation, indicating consistent ductile behavior.

Topics & Concepts

FractographyMagnesiumMaterials scienceRandom forestStructural materialGrapheneComposite materialFracture (geology)Machine learningMetallurgyComputer scienceNanotechnologyAluminum Alloys Composites PropertiesMXene and MAX Phase MaterialsMagnesium Alloys: Properties and Applications
Prediction of mechanical properties and fractography examination of AZ91 magnesium composites reinforced with graphene using a random forest machine learning model: experimental validation | Litcius