This paper presents a methodology for online Multivariate Data Analysis (MVDA) modeling techniques and gives an example of MVDA being successfully used at Iggesund Paper-board. Iggesund Paperboard (Workington) have been working in collaboration with the University of Manchester developing novel MVDA models that describe the board making process; to reduce variability in the process. This paper gives an overview of MVDA methodology and examples of how Iggesund Paper-board developed MVDA predictive model for online monitoring of bending stiffness profile. In papermaking, chemical, power and other process industries, there has been a continuing demand for higher quality products, lower product rejection rates, and the need to satisfy increasingly stringent safety and environmental regulations. Implementation and improvement of digital control schemes has been essential over the last three decades in order to meet these ever increasing standards especially since modern process plants are often large scale, highly complex, and operate with a huge number of process variables under closed loop control. Multivariate analysis methods will predominate in the future; this will result in drastic changes in the manner in which operators think about problems and how they make decisions. These methods make it possible to ask specific and precise questions of considerable complexity in natural settings. This makes it possible to conduct theoretically significant research and to evaluate the effects of naturally occurring, parametric variations in the context in which they normally occur. In this way, the natural correlations among the manifold influences on behaviour can be preserved and separate effects of these influences can be studied statistically without causing a typical isolation of either individuals or variables. Multivariate statistical techniques, including principal component analysis (PCA) and partial least squares (PLS) are capable of reducing the dimensionality of the original data such that essential information is retained; they are also able to classify data points to pre-determined classes.
展开▼
