MIXTURE PRINCIPAL COMPONENT ANALYSIS MODEL FOR MULTIVARIATE PROCESSES MONITORING
A mixture probabilistic principal component analysis model is proposed as a process monitoring tool in this paper. High-dimensional measurement data could be aggregated into some clusters based on the mixture distribution model, where the number of these clusters are automatically determined from the maximum likelihood estimation procedures. It was illustrated that the mixture PCA models conform to the multivariate data well in the experiments involving Gaussian mixtures. The multivariate statistical process monitoring mechanism is then developed first with the learning of a finite mixture model with variant principal component within each cluster, followed by the construction of the statistical process confidence intervals for the identified regions or nodes from T2 charts. For the abnormal input measurement, they would fall out of the acceptance region set by the confidence control limits.