Intelligent Parameter Screening Scheme for the AVM System

• Main Authors: Hsuan-HengHuang, 黃暄恆
• Other Authors: Fan-Tien Cheng
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/80964016859726613540

碩士 === 國立成功大學 === 製造資訊與系統研究所碩博士班 === 101 === In the past, our research team had developed the Automatic Virtual Metrology (AVM) System. This system is able to improve process capabilities and to enhance yield rates while monitoring qualities of all production pieces when implemented real-time online. However, production equipment is a dynamic and ever-changing system. The characteristics of equipment will be varied by environmental factors and throughout time period. As a result, the prediction outcomes of the current AVM system don’t seem to be accurate enough just using parameters indicated by experts. On the other hands, parameters chosen by statistical methods could be unacceptable by field process engineers due to the fact that these parameters might not have direct effects on the process and perhaps will raise monitoring cost. Therefore, in order to cost down and to improve VM accuracy, we develop an Intelligent Parameter Screening (IPS) Scheme to choose the key parameters dynamically while also consider experts’ knowledge. By applying the IPS Scheme to the AVM System, we can have better prediction results that improve performances of the AVM System.