Data based model predictive control for ring rolling

Abstract

AbstractThermomechanical ring rolling is an evolution of the process where deformation and heat treatment are combined to obtain a product with both the desired geometry and hardness or microstructure in a single step. However, the high sensitivity of the process to the initial condition and to various disturbances limits its repeatability and accuracy. In this paper, the authors implement a concept for hardness control of ring rolling in virtual experiments. A concept based on soft sensors and model predictive control is implemented on a digital twin. The operation of the different models needed for this control loop are detailed and the controller itself is illustrated.