運用LED機台製程資料建立LED亮度與電壓預測模式

國內LED產業為了維持優勢，必需提升製程技術，以高良率為生產後盾，並同時兼顧新技術產品開發及降低生產成本。LED晶粒製程中有數十個互相影響的製程參數，機台參數如果沒有通過有效的工程實驗批的測試，則又必須從頭再進行一次，耗時又耗費成本。且LED晶粒製程因資料來源多樣繁雜資料收集不易，加上無法有效率的找出產品問題，造成產品停滯在產線時間過長，則使得產品達交率差，因此如何隨時掌控生產製造過程的資訊，變成企業市場競爭力提升的關鍵因素。本研究透過MES系統收集生產批的相關資訊，首先將不同來源的資料做整合，接著使用資料前處理方式，將資料做初步處理與篩選，最後則是透過資料探勘中的類神經網路、支援向量機、迴歸分析、模式樹等分類技術，來找出最佳預測模式，其中模式樹被證實是最有效的預測技術，可以分析製程中影響亮度與電壓的有效因子，適時提供給研發、工程人員做為參考及改善依據。

關鍵字

資料探勘；良率； LED

並列摘要

In order to maintain the advantages of domestic LED industry, it is necessary to upgrade the manufacturing process not only to develop new product but also reduce production costs and improve yield. There are dozen variables which influence process parameters. If they are not been effectively test in engineering run first, often the whole process will have to start from scratch again. It's time-consuming and costly. The source of LED chip processing data are complex and hard to collect. Causing the low order fill rate because the product stuck on production line waiting for data processing. Thereby one of the key to improve competitiveness in the market is to manage process data at all time. This study using MES system to collect processing information. Staring from integrate information from different sources, and then use the data pre-processing method to do data initial processing and filtering. After, using techniques such as Neural Networks、Support Vector Machine、Regression Analysis and Model Tree of data mining to find the best predictive model. The Model Tree proved to be the best predictive model. Analysis and determine the process factor that affect brightness and voltage of LED chips. Provide engineering staff the data in time for process improvement.

並列關鍵字

Data Mining ； Yield ； LED

參考文獻

Chang, S. J., Chen, W. S., Lin, Y. C., Chang, C. S., Ko, T. K., Hsu, Y. P., . . . Shei, S. C. (2006). Nitride-based flip-chip LEDs with transparent Ohmic contacts and reflective mirrors. IEEE Transactions on Advanced Packaging, 29(3), 403-408. doi: 10.1109/tadvp.2006.871189

Liao, C. H., Chen, C. Y., Chen, H. S., Chen, K. Y., Chung, W. L., Chang, W. M., . . . Yang, C. C. (2011). Emission Efficiency Dependence on the p-GaN Thickness in a High-Indium InGaN/GaN Quantum-Well Light-Emitting Diode. IEEE Photonics Technology Letters, 23(23), 1757-1759. doi: 10.1109/lpt.2011.2169243

Alzghoul, A., Löfstrand, M., & Backe, B. (2012). Data stream forecasting for system fault prediction. Computers & Industrial Engineering, 62(4), 972-978. doi: 10.1016/j.cie.2011.12.023

An, H. M., Sim, J. I., Shin, K. S., Sung, Y. M., & Kim, T. G. (2012). Increased Light Extraction From Vertical GaN Light-Emitting Diodes With Ordered, Cone-Shaped Deep-Pillar Nanostructures. IEEE Journal of Quantum Electronics, 48(7), 891-896. doi: 10.1109/jqe.2012.2190587

Arora, A., Ceccagnoli, M., & Cohen, W. M. (2008). R&D and the patent premium. International Journal of Industrial Organization, 26(5), 1153-1179. doi: 10.1016/j.ijindorg.2007.11.004

國際替代計量

運用LED機台製程資料建立LED亮度與電壓預測模式

未授權

主題瀏覽