近年來製造業陸續投入數位化、智能化轉型，製造業是驅動經濟發展、創造就業的火車頭，智慧製造是策略目標，其中數據驅動更是核心重點，在大數據分析和人工智慧等日益成熟的新技術，可以更容易把決策者、老師傅的智慧轉化為規則或數學模式，有效結合領域知識和研判各種相關資訊 ，利用決策模型做決策讓決策模型能夠不斷地精進，使製造生產愈來愈智能化。 本研究收集機台感測器輸出訊號及人工紀錄的數據，運用資料探勘的羅吉斯回歸與決策樹方法做數據分析，讓管理者得到有價值的資訊，建立斷膜預測模型，改善人工分析的低準確率，透過預知診斷異常狀態，提早下預防對策，達到降低斷膜發生率，減少報廢金額及增加產能 。;In recent years the manufacturing industry has successively invested in digital and intelligent transformation，the manufacturing industry is the locomotive driving economic development and creating employment，smart manufacturing is a strategic goal，among which data-driven is the core focus，in the increasingly mature of new technologies such as big data analysis and man-made intelligence can easily to convert the wisdom of decision makers and teachers into rules or mathematical models，effectively combining domain knowledge and studying judging various related information，Use decision-making models to make decisions ，so that decision-making models can continuously forge ahead，making manufacturing more and more intelligent。
This research collects the output signal of the machine′s sensor and the man-made recorded data，and uses the logistic regression and decision tree methods of data exploration for data analysis，allow managers to obtain valuable information，establish predictive model of film breakage，improve the low accuracy of manual analysis，and through predictive diagnosis to diagnose abnormal conditions，early preventive measures to reduce the incidence of film breakage，reduce scrap amount and increase production capacity。
