Page 82 - Vol.45
P. 82

Tech
                                                                                                    Notes
                                                                                                    技術專文


                                                  TSMC / FACILITY PUBLISHED
           VOL.45
                                                   廠務季刊


                                                     URL.http://nfjournal/



           智能應用效率最佳化_Exhaust sub-main自動連結設備FDC即時分析系統

           Optimization of Intelligence application efficiency_Real-time analysis for exhaust
           sub-main with tool FDC


                                                                                           文│吳聖儀│台南廠務六部│


           摘要 / Abstract

              300mm廠務組織合併後,集合眾專家所長推出了廠務資訊整合平台-OneFAC,藉此輔助廠務運轉管理,也因系統整合,廠
           務有平台可獲取各廠的廠務相關資訊,可說廠務運轉管理已邁向成熟階段。然而對廠內資訊的呈現仍不夠即時,需靠著不同的
           資訊系統平台去獲取,且其資訊無法進行統計或調整,進而將廠務運轉管理持續精進。
              為提升廠務Sub-main管理品質要求,本專案整合Bluebook、InSQL、Central KER、U-chart與FDC平台,透過大數據與智能
           分析自動找出Sub-main變化與分析其原因。經資訊收集與調整,已可在Submain變化時,即時提供下游機台機況,降低83%應
           變時間,且評估設定FDC pre-Alarm,加速處理排除機台卡貨風險。

           關鍵詞 / 廠務運轉管理、系統整合、Tool FDC、智能分析

              After merger of the 300mm FAC organization, the experts in various fields launched the FAC information integration platform
           -OneFAC, which assists the factory affairs operation management. Because of the platform, we can obtain the information of each
           factory. However, the presentation of in-plant information is still not real-time enough, and it needs to be obtained through different
           system platforms. We cannot make statistics to the information, so as to continuously improve the management of FAC operations.
              In order to improve the quality requirements of submain management, this project integrates Bluebook, InSQL, Central KER,
           U-chart and FDC platforms, through big data and intelligent analysis to automatically find out the changes of submain and analyze
           the reasons. After system testing and adjustment, the status of the downstream tool can be provided in real time when the submain
           alarm occur, reducing the strain time by 83%. By setting the FDC pre-Alarm, the processing can be accelerated to eliminate the risk
           of tool jamming.

           Keywords / Facility Operation management, System integration, Tool FDC, Intelligent Analysis



           1.  前言                                            2.  文獻回顧

              目前各廠機台的運轉資訊,使用Si-view或Report整合系                    隨著工廠與廠務端監視與查詢軟體逐漸整合,資料獲取
           統進行查詢機況與相關資訊。本專案以Exhaust壓力變化為                     已不再處於被動,且隨著透明化的資訊越來越多,從中獲取相
           例,若有Exhaust相關的FDC alarm發生,廠務皆處於被動狀                關與需要的資料也相當重要。考慮將資料去蕪存菁,使用大數
           態被告知後,在進行後續的查找與詢問,耗費大量人力跟時                        據與機械學習分析,得到有用的資訊是目前最佳的方式。
           間。隨著各廠產能逐漸上升,不能停貨的壓力越來越大,若
                                                             2.1  大數據取得與解析
           能在Exhaust相關機況發生前後有預防動作,得知如何發生、
           為何發生?有助於降低因廠務相關的運轉風險。                                 「大數據」是指數量龐大而無法以傳統方式處理的資
                                                             料,無論何種產業皆能透過分析大數據預測未來趨勢,使大
                                                             數據成為各行業都在發展的數位技術。             [01]

           80
   77   78   79   80   81   82   83   84   85   86   87