Online process monitoring using a multivariate CUSUM approach with winsorization
Export citation
Abstract
With the development of Industry 4.0 (I4.0), companies are transforming the way products are designed, manufactured and distributed. The application of new technologies in production and data acquisition exacerbates the need to foster quantitative approaches in the quality management of manufactured products, such as statistical process monitoring (SPM). A measuring system machine for evaluating die-casted workpieces was designed following the previous trend. This machine already applies part of the theoretical concepts of I4.0. The presented thesis complements the application of I4.0 concepts to the device, by using SPM methods, specifically, a multivariate CUSUM to assess small and sustained shifts; where winsorizing was used to create robustness over isolated changes that can be detected using complementing Shewhart-type charts. Additionally, an online dashboard was created to display the plotting statistics in real-time.