Identification of Papermaking Process From Closed Loop Data by Bivariate Time Series Analysis

Abstract : The data used by Tee and Wu for identification of the transfer function and the noise of a Fourdrinier paper machine is reanalysed taking into account the feedback of the skilled operator. The identification procedure consists of fitting a bivariate time series to the input and output observations. The transfer function and the noise of the paper machine as well as those of the skilled operator are obtained by simple manipulations of the fitted bivariate model. The resulting model is physically interpreted and an attempt is made to locate the primary source of disturbance.