Scholarlink

<p>The application of machine learning in the manufacturing sector is gaining attention for improving both material properties and fabrication techniques. In the present study, a machine learning framework has been developed to predict the ultimate tensile strength of ceramic particles-reinforced aluminium metal matrix composites/nanocomposites fabricated via ultrasonic-assisted stir casting. Supervised machine learning models, including Elastic Net, Support Vector Regression, AdaBoost, Decision Tree, Bagging, Random Forest, Extra Trees, Gradient Boosting, and Extra Gradient Boosting, were tested and evaluated. Each ML model was trained using features such as alloy composition, reinforcement characteristics, such as particle size, density and amount, and casting parameters. The evaluation of ML models was conducted in two stages, based on <i>R</i><sup><i>2</i></sup>–values, mean absolute error, root mean square error, and deviation between predicted UTS and actual UTS of the AA7075 and its nanocomposites fabricated in this work. Among these, the random forest model was selected as the final model, which exhibited the best predictive accuracy, achieving a training <i>R</i><sup><i>2</i></sup> value of 0.982, a test <i>R</i><sup><i>2</i></sup> of 0.945, and a mean absolute error of 17.601. Bayesian optimization was further employed for hyperparameter tuning to enhance the model's generalization ability. For the final validation of the best-performing ML model, AA7075 nanocomposites were fabricated via ultrasonic-assisted stir casting using TiO<sub>2</sub> and CeO<sub>2</sub> ceramic nanoparticles as reinforcements. A minimum deviation of 0.65% between the experimental and predicted ultimate tensile strength values indicated strong agreement and confirmed the model’s capability to predict the tensile strength of the AA7075 nanocomposites.</p> Graphical abstract

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A machine learning approach to predict tensile strength of TiO2/CeO2-reinforced AA7075 nanocomposites fabricated by ultrasonic-assisted stir casting

  • 期刊论文

  • 发表日期：

    2026年4月6日

查看全文
  • Salman Ansari,
  • Sivanandam Aravindan,
  • Digavalli Ravi Kumar,
  • Ayan Bhowmik

摘要

The application of machine learning in the manufacturing sector is gaining attention for improving both material properties and fabrication techniques. In the present study, a machine learning framework has been developed to predict the ultimate tensile strength of ceramic particles-reinforced aluminium metal matrix composites/nanocomposites fabricated via ultrasonic-assisted stir casting. Supervised machine learning models, including Elastic Net, Support Vector Regression, AdaBoost, Decision Tree, Bagging, Random Forest, Extra Trees, Gradient Boosting, and Extra Gradient Boosting, were tested and evaluated. Each ML model was trained using features such as alloy composition, reinforcement characteristics, such as particle size, density and amount, and casting parameters. The evaluation of ML models was conducted in two stages, based on R2–values, mean absolute error, root mean square error, and deviation between predicted UTS and actual UTS of the AA7075 and its nanocomposites fabricated in this work. Among these, the random forest model was selected as the final model, which exhibited the best predictive accuracy, achieving a training R2 value of 0.982, a test R2 of 0.945, and a mean absolute error of 17.601. Bayesian optimization was further employed for hyperparameter tuning to enhance the model's generalization ability. For the final validation of the best-performing ML model, AA7075 nanocomposites were fabricated via ultrasonic-assisted stir casting using TiO2 and CeO2 ceramic nanoparticles as reinforcements. A minimum deviation of 0.65% between the experimental and predicted ultimate tensile strength values indicated strong agreement and confirmed the model’s capability to predict the tensile strength of the AA7075 nanocomposites.

Graphical abstract