Modeling and Control of Multiple Web Spans Using State Estimation

Abstract

Web processing systems rely on accurate lateral positioning to achieve high processing speeds and improved product quality. Due to physical constraints in some processing lines placement of the edge sensor near the web guide is not possible. As a result, large lateral oscillation and/or web instability have been observed. A new model is developed for lateral web dynamics. Experimental verification has justified the structure of the model. A parameter estimation scheme is used to tune the model for imperfections not originally incorporated. State estimation is then used to predict lateral web position on a downstream sensor. Desirable control is achieved and is further improved with the use of the feedforward sensor. Predicted and experimental results are compared.