Statistical monitoring of multi-stage processes

Abstract

In many complex processes, such as semiconductor manufacturing or production of mass storage systems, a large number of variables are monitored simultaneously. These variables can typically be impacted by several points of the manufacturing process, necessitating efforts that include not only monitoring but also diagnostics that includes establishing change-points, regimes and potential stages of influence. We discuss statistical methods used to handle such multi-stage data and give examples of applying these methods in large-scale monitoring systems.