Gurunath V Shinde, Abhijeet Suryawanshi, Niranjana Behera
{"title":"Prediction of friction stir welding performances of dissimilar AA3003-H12 and C12200-H01 using machine learning algorithms","authors":"Gurunath V Shinde, Abhijeet Suryawanshi, Niranjana Behera","doi":"10.1177/09544089241272824","DOIUrl":null,"url":null,"abstract":"Tests specimens were prepared by friction stir welding of two dissimilar metals aluminum and copper. The specimens were subjected to mechanical tests to calculate the ultimate tensile strength, yield strength, percentage elongation, and impact energy. Four different machine learning algorithms (AdaBoost, CatBoost, Gradient Boosting, and XGBoost) were applied for developing the ML models in predicting the performance parameters such as ultimate strength, yield strength, percentage elongation, and impact energy. Pin type, weld speed, rotational speed, and shoulder diameter were considered as the input parameters for the model. Training, testing, and validation were carried out by considering 60%, 20%, and 20% of the available data respectively. In terms of accuracy (lower MAE, lower RMSE, greater R<jats:sup>2</jats:sup> value, and lower AAD%), CatBoost model, Gradient Boosting model, and XGBoost model performed better than the AdaBoost model in predicting the ultimate tensile strength, yield strength, percentage elongation, and impact energy. Compared to other models, AdaBoost model has only few hyperparameters for fine-tuning. During hyperparameters tuning, AdaBoost model showed accuracy only within a narrow range of values of features.","PeriodicalId":20552,"journal":{"name":"Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering","volume":null,"pages":null},"PeriodicalIF":2.3000,"publicationDate":"2024-09-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.1177/09544089241272824","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, MECHANICAL","Score":null,"Total":0}
引用次数: 0
Abstract
Tests specimens were prepared by friction stir welding of two dissimilar metals aluminum and copper. The specimens were subjected to mechanical tests to calculate the ultimate tensile strength, yield strength, percentage elongation, and impact energy. Four different machine learning algorithms (AdaBoost, CatBoost, Gradient Boosting, and XGBoost) were applied for developing the ML models in predicting the performance parameters such as ultimate strength, yield strength, percentage elongation, and impact energy. Pin type, weld speed, rotational speed, and shoulder diameter were considered as the input parameters for the model. Training, testing, and validation were carried out by considering 60%, 20%, and 20% of the available data respectively. In terms of accuracy (lower MAE, lower RMSE, greater R2 value, and lower AAD%), CatBoost model, Gradient Boosting model, and XGBoost model performed better than the AdaBoost model in predicting the ultimate tensile strength, yield strength, percentage elongation, and impact energy. Compared to other models, AdaBoost model has only few hyperparameters for fine-tuning. During hyperparameters tuning, AdaBoost model showed accuracy only within a narrow range of values of features.
期刊介绍:
The Journal of Process Mechanical Engineering publishes high-quality, peer-reviewed papers covering a broad area of mechanical engineering activities associated with the design and operation of process equipment.