Hybrid machine learning model with optimization algorithm for predicting the incubation dose of void swelling in irradiated metals

Abstract

This study introduces novel hybrid machine learning (ML) models that integrate six state-of-the-art ML algorithms with the Harris Hawks Optimization (HHO) algorithm to enhance the prediction of the incubation dose in irradiated metals. A comprehensive database comprising 305 experimental samples with 24 input features is used to develop the models, with hyperparameters optimized through a combination of cross-validation method and HHO. Performance evaluation across various metrics identifies the hybrid model combining HHO and categorical gradient boosting (CGB), named HHO-CGB, as the most accurate and stable for predicting the incubation dose. To gain further insights, the Shapley Additive Explanations method is employed to assess the global and local contributions of input variables, revealing Fe (wt.%), temperature (K), dose rate (dpa/s), and V (wt.%) as the most influential factors. Finally, a user-friendly graphical interface tool and web application are developed based on the HHO-CGB model, providing a practical and cost-effective solution for predicting the incubation dose of irradiated metals.