A two-step multivariate statistical learning approach for batch process soft sensing

• DOI: 10.1016/j.dche.2021.100003
• Publication Date: 2021-10-19T16:32:27Z
• Authors: Aaron Hicks; Matthew Johnston; Max Mowbray; Maxwell Barton; Amanda Lane; Cesar Mendoza; Philip Martin; Dongda Zhang
• Keywords: Soft sensor; Factory (object-oriented programming); Statistical Process Control

Statistical machine learning algorithms have been widely used to analyse industrial data for batch process monitoring and control. In this study, we aimed take a two-step approach systematically reduce dimensionality design soft-sensors product quality prediction. The first employs partial least squares screen the entire dataset identify critical time regions operational variables, then adopts multiway construct within reduced space estimate final quality. Innovations of include ease visualisation ability major activities factory. To highlight efficiency practical benefits, an personal care manufacturing was presented as example two were successfully developed end viscosity estimation. Furthermore, accuracy, reliability, thoroughly discussed. This paper, therefore, demonstrates potential proposed approach.