Diseño de una carta de control basada en Análisis de Componentes Principales : un caso de estudio

The set of quantitative methods and techniques used to detect assignable variations in manufacturing processes are contained within a discipline classified as statistical process control. Such methods carry out precise evaluations on the general state of productive systems and carry out simultaneous monitoring of various interrelated quality characteristics. In the framework of this research, the analysis of a chemical process based on the theoretical principles of principal component analysis is proposed, which enables the representation of the original variables in a compact dimensional space. In the later phase, a control graph based on the squares of the prediction errors is constructed in order to evaluate the behavior of the composite variables found. The results indicate that the process is not marginally stable and it is necessary to reduce its variability margin.