Assessing the Temperature-Dependent Characteristics of Warm-Rolled EN8 Steel through Magnetic Barkhausen Noise Technique and Predictive Modeling with Voting Regression Technique

This study investigates the warm rolling of EN8 steel at temperatures of 600 °C, 700 °C, and 800 °C with varying passes (10,15,20). Post-rolling, microstructural analysis and mechanical property evaluation were performed. Magnetic Barkhausen noise signals were collected, and features like peak value and RMS value were extracted. A voting regression-based machine learning model was used to predict hardness and Charpy strength. The model achieved R² scores of 0.912 for training and 0.894 for testing in predicting hardness, and 0.868 for training and 0.815 for testing in predicting Charpy strength. The normal distribution of residuals for Charpy strength indicates accurate predictions. The present work highlights the applicability of magnetic Barkhausen noise measurements for nondestructive evaluation of microhardness and fracture toughness, benefiting maintenance and reliability in construction and transportation. Future research should focus on nonlinear models and automatic feature selection to improve prediction accuracy.