Summary: | 碩士 === 靜宜大學 === 資訊碩士在職專班 === 98 === LCD technology had significant improvement today and highl competition in the industry. The most important to lead the head is not only to delivery new product quickly with best quality and reliability certification, but also to have cost efficiency, The purpose of this paper is to provide the TFT LCD reliability and burn-in testing time assessment model. It can help to reduce production cycle time for mass production that help the company to get most profit.
The research is to explain how to analyze the infant mortality of TFT-LCD modules by data mining. After analyzing the data, the decision tree of data classification can be applied to build up the prediction model. It can help us to predict the infant mortality of TFT-LCD modules, and find the failure of product rapidly. Therefore, the burn In and production time of TFT-LCD modules can be shortened, the quality and reliability of products would also be kept, to create the more profit.
We can help to improve T420HW04-0/01and T460HW03-1/00 tool type burn-in evaluation time from 12 hours to 3 hours by decision tree model result with MTBF large than 60000 hours ( 90% confidence level).
Meanwhile, we can use this paper model to provide a unified burn-in assessment methodology of each Fab for LCD production evaluation.
|