DYNAMIC INFERENTIAL ESTIMATION USING PRINCIPAL COMPONENTS REGRESSION (PCR)

Principal components regression (PCR) is applied to the dynamic infere ntial estimation of plant outputs from highly correlated data. A genet ic algorithm (GA) approach is developed for the optimal selection of s ubsets from the available measurement variables, thereby providing a m ethod of identifying nonessential elements. The theoretical link betwe en principal components analysis (PCA) and state-space modelling is em ployed to identify a measurement equation involving the GA-selected su bset, which is then used for inferential estimation of the omitted var iables. These techniques are successfully demonstrated for the inferen tial estimation of outputs from a validated industrial benchmark simul ation of an overheads condenser and reflux drum model (OCRD). (C) 1998 Elsevier Science B.V. All rights reserved.