Dual-driven predictive model for wear mechanism of work rolls in thin slab casting and rolling production line

The work rolls in modern thin slab continuous casting and rolling (TSCR) lines operate under both high speeds and heavy loads, making the real-time prediction of uneven roll wear a significant technical challenge. Existing prediction methods often fail to effectively integrate physical wear mechanisms with data-driven approaches. This study presents a novel predictive method that couples production data with mechanistic wear modeling. The proposed model incorporates rolling parameters, temperature effects, and deformation mechanics, and employs particle swarm optimization (PSO) along with segmented solving techniques to introduce adaptive weight factors. Validation results demonstrate the model achieves high accuracy and strong adaptability across production lines, offering a promising framework for wear prediction in TSCR lines and potential applications in other industrial domains.