Developing statistical models in an early warning system and its empirical study

Abstract

When a new equipment or process is released, it is critical to ensure it behave as expected and stay in normal condition. The study proposes a research framework in which a statistical model is constructed for newly released equipment and process monitoring. An empirical study is conducted in a DRAM fabrication facility for validation. Based on the model, a best set of sample test items which discriminates the newly released equipment is selected and a group of normal equipments is obtained. Thus, the alarm signals can be triggered in an early warning system.