LED模組及光學干涉技術於TFT-LCD系統之應用發展;Development of LED Module and Optical Interference Technology for TFT-LCD Applications

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Electrical Engineering > Electronic Thesis & Dissertation > Item 987654321/59246

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/59246

Title:	LED模組及光學干涉技術於TFT-LCD系統之應用發展;Development of LED Module and Optical Interference Technology for TFT-LCD Applications
Authors:	李東諺;Li,Tung-Yen
Contributors:	電機工程學系
Keywords:	干涉光學;液晶面板;背光模組;類神經網路;backlight module;interference pattern;LCD panel;neural network
Date:	2013-01-18
Issue Date:	2013-03-25 16:18:34 (UTC+8)
Publisher:	國立中央大學
Abstract:	TFTLCD顯示器為目前平面顯示器中的主流，TFTLCD顯示器主要是由液晶面板模組，背光模組以及電路驅動模組三個部份所構成。本篇論文提供了TFTLCD組件中有關新型背光模組的設計以及液晶面板的檢驗方式。我們使用LED模組為背光光源同時搭配微反射結構和光學擴散元件進行新型直下式背光模組的設計。經過實際測量，我們的設計相對於2003年R.S.West的設計，在均勻度上我們增加了24%（由60%到84%），亮度上我們增加了23.29%(由10000nt到12329nt)。另一方面，本論文提出了應用光學干涉的方法搭配類神經網路及統計分類方式進行TFTLCD面板中有關Gap Mura Defect的檢驗分類。我們目前可以將固定膠過多，面板內有外來物以及固定膠內有纖維聚集這3種造成Gap Mura Defect的原因檢出。搭配使用Neural Network Classification方法，Mean Squared Error (MSE)經過反覆學習後可以降低至0.01以下。由此方式可以將製程中不良的面板提前檢出，減低製作不良品的機率，藉以提高產線的良率。Recently, TFT-LCD is the most popular display application in the flat panel display application. The main parts in the TFT-LCD system includes: LCD panel module, backlight module and electronic driver module. This paper provides the novel design of direct-in backlight module of the TFT-LCD system and the inspection method for the gap mura defect of the LCD panel. We applied the LED light source, secondary light-guide component and micro reflector structure to design the novel backlight module. Compared with West’s work in 2003, the uniformity ratio for our new design shows an increase of 24%, from 60% to 84%, and the luminance a 23.29% improvement, from 10 000 nits to 12 329 nits. On the other hand, we combine the optical interference method and neural network classification to detect the gap mura defect of the LCD panel. Three kinds of mura defects including non-uniformity sealant panel; panel with foreign material and fiber-cluster panel are tested. After learning process of Neural Network Classification method, the Mean Squared Error could decrease less than 0.01. By this method, we could sort out the bad panel and increase the yield rate of the production line.
Appears in Collections:	[Graduate Institute of Electrical Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	840	View/Open

社群 sharing

Loading...