Tensile Strength Prediction of Gray Cast Iron for Cylinder Head Based on Microstructure and Machine Learning

The ultimate tensile strength (UTS) of gray cast iron (GCI) can be affected by numerous parameters due to its complex microstructures. To further understand the UTS of GCI, it is necessary to evaluate the impact of various parameters. Herein, a UTS prediction method based on microstructure features and machine learning (ML) algorithms is proposed. The six regression algorithms, namely, Bayesian Ridge, Linear Regression, Elastic Net Regression, Support Vector Regression, Gradient Boosting Regressor (GBR), and Random Forest Regressor are used to develop the prediction models. The predicted results show that the GBR has the best prediction performance for the predicted UTS and the error bands within 5%. The feature importance indicates that matrix hardness has the greatest effect on the UTS in the ML models. Several machine learning algorithms are used to evaluate the tensile strength of metals based on microstructure characteristics. These models can accurately predict the tensile properties of gray cast iron and rank the importance of the microstructural features referenced in the models, which can guide the application of machine learning algorithms in tensile prediction and alloy design of gray cast iron.