Calendering Process MPC using recursive DMDc

Ensuring high quality of various layers within a tire demands efficient process control. Rubber calendering does not depart from this important evolution of the tire machinery world. Among the numerous variables and parameters involved in calendering, specific attention is given here to the model predictive control of the rubber temperature. This focus is driven by two key factors: the highly nonlinear dynamics of rubber temperature and the stringent temperature constraints inherent in manufacturing processes. Our rubber temperature control solution combines a constrained MPC algorithm with a time-varying model generated using a recursive form of the well-established Dynamic Mode Decomposition with control (DMDc) method. As demonstrated subsequently, this online approach produces a reduced-order, time-varying state-space representation capable of accurately approximating the nonlinear dynamics of rubber temperature. The entire algorithm has been successfully tested using simulated data derived from a high fidelity simulator replicating a calendering process.