EXPONENTIALLY WEIGHTED MOVING PRINCIPAL COMPONENTS-ANALYSIS AND PROJECTIONS TO LATENT STRUCTURES

For stable (non-dynamic) chemical processes characterized by multivariate data, principal components analysis (PCA) and projections to latent structures (PLS) have recently been shown to provide useful monitoring schemes. In this work, PCA and PLS are generalized to dynamically updated models for modelling processes with memory and drift. These models are based on exponentially weighted observations, and are formulated as multivariate generalizations of the exponentially weighted moving average (EWMA). Principles and estimation algorithms for EWM-PCA and EWM-PLS are presented, and predictive control schemes based on these models are discussed.