The application of dynamic principal component analysis to enhance chunk monitoring of an industrial fluidized-bed reactor

Dynamic principal component analysis(DPCA) is an extension of conventional principal component analysis (PCA) for dealing with multivariate dynamic data. We adopted DPCA to enhance chunk monitoring of an industrial fluidized-bed reactor, overcoming the shortcomings of conventional monitoring schemes. The appropriate methods for application of DPCA were proposed, such as combining parallel analysis and Akaike information criterion (AIC) or Bayesian information criterion @IC) to determine the number of lagged variables, using empirical reference distribution (Em) based non-parametric control limits for the statistics which do not follow theoretical distribution, and adopting data smoothing to reduce the idluence of the noises. A DPCA model for chunk monitoring is constructed using the industrial data and its effectiveness is verified.