透過您的圖書館登入
IP:3.16.218.62
  • 學位論文

印刷電路板工作站內製程品質資料挖掘及建構異常診斷系統

he Development of Internal Process Quality Handling Model for Printed Circuit Board Industries Using Data Mining Techniques

指導教授 : 鄭春生 博士
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


當前電子產業興盛,相關產業以「降低生產及營業成本」、「追求卓越品質」作為企業與企業之間競爭之績效指標。對企業而言,無論是對「內部顧客」或「外部顧客」,都必須提供滿足需求、符合規格、品質良好之產品。內部製程品質控制,往往與人員、機器設備、方法、材料、控制參數等因素有關。當品質異常發生時,製程及品保工程人員應用矯正行動處理單 (corrective action request, CAR),進行製程工作站之異常診斷分析,藉由正確之分析及改善,消除變異來源,繼而提供良好品質之產品予下製程或最終顧客。 本研究主要針對印刷電路板業,提出一個工作站內製程品質異常診斷系統。應用資料挖掘技術,將矯正行動處理單,轉換成資料分析所用之資料形式,挖掘製程品質異常之重要肇因。先以「屬性及欄位」定義,將矯正行動處理單上工程人員所分析之變異來源與異常狀況,整建成各工作站資料庫。再以決策樹分析,將變異來源與異常狀況予以分類歸納、產生法則。本研究評估資料挖掘分析工具之總平均準確率,作為工具選用依據。針對歸納法則,藉由「專家評估表」方式,評估法則之正確實用性。由實際案例研究,驗證本研究有助於製程品質異常診斷。

並列摘要


In today’s electronic industry flourish, all enterprises is devoted to cost down and fabricated for best quality, to be the effectively competitive index. In the enterprises, no matter internal customer or external customer all should be provided satisfactions, fit the specification, good quality product for both of two. Internal quality control always relates to manpower, machine, method, material and controlled parameter. Once quality issue occurred, production and quality engineers will issue the corrective action request (CAR) for analysis. Through the analysis improve the quality and eliminate the defect cause to offer the product with good quality to the customer. The research focus on printed circuit board (PCB) industry, bring up the internal process quality diagnosis information system. The research utilizes two data mining defective analyzed models (See5 and PolyAnalyst) to build the diagnosis knowledge database to address the root causes of defective phenomenon by using data from internal CAR. Define the attribute and column, transfer the information to data base. Utilize the decision tree to classify the data base and transfer to be the rules. The research evaluate two data mining defective analyzed models and the results shown the performance of See5 is better than that of PolyAnalyst in the case. In addition to evaluate the correct and practicability of the rules, use the expert evaluation form for analysis. The results from the 5 PCB manufacturers demonstrate that the rule is a useful tool for diagnosis defect mode.

參考文獻


14. 劉中光,「以資料挖掘為基建構製程品質問題診斷系統─以印刷電路板業為例」,元智大學工業工程與管理研究所碩士論文,2001
1. 丁振原,「以資料探勘技術為基建構PCB客訴問題處理模式」,元智大學工業工程與管理研究所碩士論文,2002。
15. 蔡智政,「應用CART決策樹與資料視覺技術於低良率晶圓成因探討」,元智大學工業工程與管理研究所碩士論文,2001。
17. Barlow, T. and P. Neville, “Case study: Visualization for decision tree analysis in data mining,” IEEE Symposium on Information Visualization, San Diego, CA, 149-152 (2001).
18. Berry, M. and G. Linoff, Mastering Data Mining, the Art & Science of Cutomer Relation Management, John Wiley, New York (1999).

被引用紀錄


周雅君(2007)。以資料探勘為基建構偏光板品質異常診斷系統〔碩士論文,元智大學〕。華藝線上圖書館。https://doi.org/10.6838/YZU.2007.00195
謝惠如(2010)。拋光製程破片關鍵影響因子辨識模型〔碩士論文,國立清華大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0016-1901201111393211

延伸閱讀