Defect clustering indentification in color filter fabrication process using data mining techniques
碩士 === 國立成功大學 === 工業與資訊管理學系專班 === 95 === The name of TFT-LCD is the abbreviation of "Thin Film Transistor Liquid Crystal Display " TFT-LCD consists of many parts. Colored filter is the most costly module. This research investigated the defects consecutive clustering phenomenon occured in...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2007
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/28017352728640695914 |
| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系專班 === 95 === The name of TFT-LCD is the abbreviation of "Thin Film Transistor Liquid Crystal Display " TFT-LCD consists of many parts. Colored filter is the most costly module. This research investigated the defects consecutive clustering phenomenon occured in colored filter fabrication process. The phenomenon can not be controlled effectively by current SPC control charts. The purpose of this research is to locate defective zones promptly from defected data using data mining. Firstly, the "decision tree" is used to find out the most possible defected process. Secondly, the defected data collected from the identified process is clustered using k-mean method. Finally , we validate the defected clustering phenomenon by visualization techniques. The quality of TFT - LCD will be improved if the causes of defected clustering phenomenon can be analyzed in a short
time.
|
|---|