Multivariate Pattern Recognition in MSPC Using Bayesian Inference

Multivariate Statistical Process Control (MSPC) seeks to monitor several quality characteristics simultaneously. However, it has limitations derived from its inability identify the source of special variation in process. In this research, a proposed model that does not have limitation is presented. paper, data two scenarios were used: (A) created by simulation and (B) random variable obtained analysed product, which case corresponds cheese production slicing process dairy industry. The includes dimensional reduction procedure based on centrality dispersion. goal recognise multivariate pattern conjunction univariate variables with patterns so indicates pattern. consists stages. first stage concerned identification uses Moving Windows (MWs) for segmentation analysis. second Bayesian Inference techniques such as conditional probabilities Networks. By using these techniques, contributed found obtained. Furthermore, evaluates probability individual generating specific variable. This interpreted signal performance allows out-of-control state causes failure. efficiency results compared favourably respect Hotelling’s T2 chart, validates our model.