Franco Colleoni, Tomislav Vucina, Claudio Araya, Jacques Sebag, Vera C. Rubin Observatory, La Serena, Chile
The Vera C. Rubin Observatory Coating plant consists of a Coating Chamber for highly reflective coating deposition, using sputtering process with planar magnetrons. As the coating process in general consists of several independents and also correlated variables, it’s highly desirable to have mathematical models that can predict or forecast several process variables that allows to improve our coating process, planning and performance.
Going from simple regression models to neural networks, this paper will show some models used to predict different phenomenon and processes like a) target degradation (to predict when to change targets instead of relying on visual inspection), b) virtual thickness sensors (to model film thickness on places when sensors can’t be installed), c) virtual plant model (to integrate all process variables and characterize the degree of influence of the different variables for the coating process) and others.