Asymmetric deformation modelling and fast adaptive MPC for slab camber control in hot rough rolling

Slab camber is a critical shape-defect problem in hot strip rough rolling and directly affects slab centring, side-guide interaction and subsequent rolling stability. Excessive slab camber is especially problematic during transient threading stages, particularly at the head end, where asymmetric rolling conditions are more pronounced. To address this problem, this article proposes a physics-guided analysis and fast control method based on asymmetric deformation modelling. A mathematical model of slab camber motion is established by considering the combined effects of incoming wedge, initial curvature, centre offset, and stiffness difference. Based on this model, the influences of different asymmetric factors on exit wedge evolution and transient lateral motion are quantitatively analysed in an interpretable manner. Accordingly, an auto-tuning method for fast model predictive control (F-MPC) with an adaptive time horizon is proposed to dynamically suppress slab camber. The optimal time-horizon combination is obtained by particle swarm optimisation (PSO) for offline optimisation, and the online computational burden is reduced by an interior-point method with warm-start initialisation to satisfy control-cycle requirements. Simulations and mill trials show that the F-MPC controller effectively reduces slab camber at the roughing-mill exit and is suitable for industrial online implementation.