半導體廠生產電力品質管理之研究

Abstract

電力系統對於半導體產業而言，是製造生產的中樞神經。一旦電力系統發生故障或供電品質異常，將會引起生產線的機台跳機，而中斷生產流程，且隨著復電時間的延長，時間與金錢的損失上也就愈來愈大。因此，半導體產業對於電力系統品質的好壞相當注重。為了有效監控廠區電力系統品質與異常的變化，本篇論文建構多變量統計製程管制法，來監測電力系統異常。透過所建立的多變量管制圖之監控模式，可在95%信賴區間下，可以有效監控與預警電壓品質異常之變化，以有效的警報發佈，降低誤警次數，提供現場工程師緊急應變處理之判斷，提升工程師的工作效率。

Electric System is like as central nervous system for hi technology industries. As electric system is broken down, it could be leade to the interruption of production line. However, with increasing recovery time of electric system, the cost of time and money becomes huge problems. In order to monitor the change of electric system, this paper utilizes the Mutvariable Static Process Control method ( T-Square, SPC ) to predict the qualities of electric voltage. These considered factors refer to Busbar, TR1, TR2, TR3, FAB1 of VA, VB and VC. This paper also sets up MSPC of monitoring model with 95% confidence level. It is not only can supervise and prewarn the changes of voltage, but also reduce the probabilities of fault detection. On site engineers can effectively achieve the fault treatment of electric system.