Identification tool for cross-directional processes

Abstract

This paper considers an industrial identification tool for cross-directional (CD) process of continuous web manufacturing such as papermaking. A special focus is on identification of the mapping between CD actuators and measured profiles of the web properties from input-output process data. The developed algorithms are based on nonlinear parametric models of the CD response shape and mapping and minimize a model fit error for a two-dimensional array of data. The algorithms are deployed as a part of an automated process control support tool and have been successfully used over a period of time on many paper mills with various types of the CD actuators. The paper is illustrated by identification results from a real-life paper mill.