Cumulative Thermal Roll Profile Effect of Multi-Stand Zone Cooling in Hot Strip Rolling and Collaborative Regulation Method for Strip Crown

The strong coupling between the dynamically evolving thermal roll profile and strip shape across a multi-stand hot finishing mill poses a major challenge to shape control. To address this issue, this paper proposes a multi-stand collaborative zone cooling method based on a thermo-mechanical coupling inverse problem. A dynamic heat-transfer coefficient model incorporating multi-source disturbances (spray angle, water flow density, and roll surface temperature) is established, and an online inversion based on the Gauss–Newton method using measured post-service roll temperatures improves the temperature prediction accuracy to over 95%. A cascade-inheritance mechanism is revealed, whereby the thermal crown formed at an upstream stand alters the incoming strip profile and thereby affects the roll gap of downstream stands; the associated decrease in rolling-force regulation efficiency from F1 to F7 is explained by the increasing strip plastic stiffness. A differentiated strategy is accordingly developed: upstream stands (F1–F3) actively regulate lateral metal flow through the thermal crown, whereas downstream stands (F4–F7) rely on zone cooling to compensate for plastic stiffness effects. A mapping from the loaded roll gap profile to the zone cooling flow is established, and a Mamdani fuzzy adaptive algorithm provides real-time compensation within the 150 ms control calculation cycle (distinct from the up-to-10 s physical response time of the actuators). Industrial validation on the downstream stands (F5–F7) of a 1580 mm hot strip finishing mill reduced the upper/lower work roll temperature difference from 7 to 12 °C to within 5 °C, decreased the average C40 crown by 8.7%, raised the excellent edge rate (C15–C40 ≤ 20 µm) from 59.4% to 83.4%, and expanded the adjustable range of the thermal crown by more than 2.5 times. In addition, numerical analyses extend the collaborative regulation strategy to the full mill (F1–F7) and demonstrate its applicability; these full-mill results are model-based predictions rather than directly validated industrial responses.