A novel intelligent method for slab front-end bending control in hot rolling

A novel intelligent control strategy based on machine learning (ML) and an optimal feed-forward control method are proposed to realize the high-precision prediction and precise control of slab front-end bending in hot rolling. By analyzing the mechanism of slab front-end bending, the factors influencing hot rolling slab front-end bending are analyzed by a simulation model, and the actuator efficiency model of slab front-end bending is established. Through exploring the genetic law of slab front-end bending, an optimization design for slab front-end bending control that combines bar feed-forward with pass feed-forward is presented. By comparing different ML methods, the XGBoost model is selected to establish a prediction model and is combined the proposed optimal control strategy to realize slab front-end bending control. The proposed control strategy has been successfully applied to industrial sites and achieved satisfactory results, which the proportion of the serious slab turn-up phenomenon has been reduced by 11.12%, and the proportion of bending values within 50 mm has increased by 18.56%. The product quality is significantly improved, and automatic control of slab front-end bending for hot rolling is realized.