Corrosion resistant performance prediction in high-entropy alloys: A framework for model, interpretation and multi-dimensional visualization

This study proposes a multi-model ensemble-based framework for High-entropy alloy (HEA) corrosion resistance prediction and interpretation. The NRKG-S model first predicts the crystal structure based on composition and processing, followed by corrosion resistance prediction. The Cross-Validation Model-Based Integrated Prediction and High-Dimensional Interpretative Visualization Methods explore the full compositional and processing space, providing insights into the relationships between composition, processing, and corrosion resistance. Experimental results show the framework accurately predicts HEA corrosion resistance, and the partial conclusions regarding model interpretability align with existing studies, offering new perspectives for material design, optimization, and understanding machine learning models.