Prediction of the swelling rate of irradiated type 316 stainless steels via machine learning methods

•The machine learning (ML) model which predicts the swelling rate of irradiated type 316 stainless steels is developed;.•The learned model can predict the irradiation-induced swelling rate of type 316 stainless steels with reasonable accuracy;.•The most controlling factors identified by the ML model are in agreement with the established understanding of factors affecting the irradiation-induced swelling rate of type 316 stainless steels. Irradiation-induced void swelling typically leads to the degradation of mechanical properties of metals and the significant decrease in material density, which influences the safety of nuclear reactors. It is essential to understand the irradiation-induced swelling behavior of alloys. In this work, machine learning (ML) methods are adopted to make a prediction on the swelling rate of type 316 stainless steels. Based on the defined target variable and potentially influential factors, the dataset which contains 333 samples and 23 features is created with excavating previous studies. The correlation analysis between all pairs of variables is performed firstly. Based on the grid search technology, k-fold cross validation is used to obtain the optimal combination of parameters for ML models. Five ML models are fitted and their predictive performances are compared. Because GradientBoosting (GB) model exhibits the best prediction, it is used to identify the most controlling factors in the prediction of the irradiation-induced swelling behavior of alloys. The GB model gives the results which are in agreement with the established conventional understanding of factors affecting the swelling rate of type 316 stainless steels. As a summary, the application of ML to the prediction of the irradiation-induced swelling behavior of alloys provides a new insight to the development of irradiation resistance steels.