Model predictive control of CD profiles in sheet forming processes using full profile disturbance models identified by adaptive PCA

In this paper a method to control full cross-direction (CD) profile variations in sheet forming processes is presented. The Karhunen-Loeve (KL) expansion, a particular principal components analysis (PCA) technique, is used to construct a disturbance profile model from full CD profile measurements. It is shown how to build a time-varying disturbance state-space model using the KL expansion and low order autoregressive processes. This disturbance predictor is then incorporated in a state-space model predictive control algorithm which includes constraints on the actuators. Several paper machine closed loop simulations are presented to show how the method works for different types of actuator constraints.