Proceedings of the Institution of Mechanical Engineers Part E Journal of Process Mechanical Engineering, Journal Year: 2025, Volume and Issue: unknown
Published: March 25, 2025
An integrated machine learning system has been developed to make creep strain predictions on shape-memory Nitinol alloys subjected different stress ranging from 0 500 MPa and temperature ranges 25°C 80°C. Our work used experimental data 35 strain-controlled tests through which we evaluated four approaches, including linear regression k-nearest neighbours along with decision Tree Random Forest because traditional methods showed insufficient capabilities for analysing stress/temperature/creep deformation relationships. The evaluation of the models occurred mean absolute error(MAE) root square error (RMSE) coefficient determination ( R ²). Ensemble Method exceeded performance Decision by providing near-perfect predictive power ² = 1.000 while maintaining a zero MAE 0.000 RMSE 3.93 × 10 9 . This matched better than (e.g. regression: 0.708, 0.009) tree 0.999, 0.000). Stress exists as main cause based observation there is positive correlation r > 0.95) maintains non-linear effect that hastens rates 40% at 80°C when compared 25°C. study leads field merging algorithms thermomechanical build comprehensive increases material reliability optimisation potential biomedical aerospace energy sectors.
Language: Английский