應用滾動灰預測模型之SPC-EPC製程回饋控制系統

Abstract

在產品的製造過程中，常常存在一些人為的干擾(disturbances)或一些無法避免的自然干擾，而使製程發生變異。為了提升產品之競爭力，製造商必須持續地降低製程變異，以提昇產品品質。統計製程管制(Statistical Process Control, SPC)及工程製程管制(Engineering Process Control, EPC)是業界常用來監控製程變異的手法。SPC主要是運用管制圖來找出造成製程變異的可歸屬原因，而EPC則是運用回饋控制與參數調整的方式，使製程產出值維持在目標值附近。在整合SPC與EPC的架構下，除了可偵測製程是否有可歸屬原因干擾外，同時可使製程產出值保持穩定，皆比其單獨使用SPC或EPC為佳。本研究主要之目的是利用產品製造過程中生產機台與量測機台產生之製程數據，發展一套結合SPC與EPC的製程資訊回饋控制系統，以有效且即時地降低製程之變異。在回饋控制系統中，本研究首先利用逐步迴歸法(Stepwise Regression Analysis)將影響製程的顯著參數依其重要性排序，同時建構產出值與製程參數間的關係式。當產出值明顯偏離目標值，但未違反SPC管制規則時，則依產出值與製程參數之關係式即可快速提供應調整的製程參數及調整之幅度。此外，本研究建構一滾動灰預測模型，在製程產出值呈現非隨機的趨勢時，可以預測未來時間點之製程產出值，以便回饋控制系統能快速準確地事先決定是否須啟動製程回饋與參數調整機制，以有效監控及降低變異。

In order to enhance the competitiveness of products, manufacturers must reduce process variation to improve the product quality. Statistical Process Control (SPC) is frequently employed to remove the assignable cause and Engineering Process Control (EPC) is also implemented to adjust the process input, so that the process output can be revised quickly to the target value. Many studies recommended that integrated SPC-EPC has better performance than just employing SPC or EPC alone. The main purpose of this study is using the data of the production and metrology tools to develop a framework of SPC- EPC process feedback control system. First, this paper uses the stepwise regression analysis to order the significant input variables according to their importance, then constructing a relationship between the process input and the output. In addition, a rolling gray model is constructed to predict the future process output when the output of the process has a non-random trend. This feedback control system can decide in advance whether it is necessary to start the variable adjustment mechanism to effectively reduce the process variation.