From Two-Dimensional Linear Quadratic Optimal Control to Iterative Learning Control. Paper 2. Iterative Learning Controls for Batch Processes

On the basis of the two-dimensional linear quadratic (2DLQ) optimal control developed in the first paper of this series, systematical design and analysis of optimal iterative learning control (ILC) schemes are developed in this paper. The structure analysis indicates that the resulting optimal ILC laws are the combination of 2D ILC and 2D state feedback control optimized in a 2D sense. The proposed 2D based design methods not only result in a unified design of these two types of control but also allow easy extension of 2D information for better/balanced performance in terms of both time and batches. To avoid state measurement/estimation, a state-space realization based on input-output information is presented. The applicability and effectiveness of the proposed schemes are illustrated with velocity control in injection molding.