Model Predictive Control of a High Power Rolling-Mill Drive Considering Shaft Torque Constraints

Model Predictive Control enables the multi-objective optimization of mechatronic drive systems including the introduction of shaft torque constraints. In this paper, its practical application to a high power drive is discussed. This includes the controller design procedure, that particularly considers the limited torque dynamics of the controlled drive motor and state-estimation errors, that are counteracted by the implementation of soft constraints. As compromise between modelling errors and computational effort, the drive train is approximated as a three-mass model for state estimation and prediction. Simulative investigations indicate, that additional measures must be taken to achieve optimal results as the system performance is severely impacted by the observer’s dynamic reaction on plant perturbations and disturbances. Compared to a PI controller, the Model Predictive Control leads to reduced mechanical stress by decreasing the maximum shaft torques and suppressing the oscillations.