An Anomaly Detection Method for Fluorine Discharge in Semiconductor Tools

• Main Authors: J-K Hsu, 許景凱
• Other Authors: 廖宜恩
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/34928553570818134187

碩士 === 國立中興大學 === 資訊科學與工程學系 === 104 === As the environmental awareness has increased in the society, environmental issues become more important for enterprise social responsibility. In semiconductor industries, the efficient treatment of exhaust gas, especially Fluorine discharge F2, produced by machines is highly related to the environmental safty and pollution prevention. For finding anomalous exhaust emission, engineers usually monitor the records of exhaust gas from machines and then find anomaly machines manually. However, these conventional methods cannot efficiently find the abnormal machines tp precent air pollutant from discharge. In this thesis, we uses all of the records from LSC and the exhaust pipe system to perform the associaion rule mining. By using the association rules and the frequent itemsets, abnormal machines can be found efficiently. The experimental results show that the proposed methods have accuracy of 80%, 80%, 90%, and 94% for finding abnormal fluorine discharge machines using the sensors on the floors of B1F, 1F, 2F, and 3F, respectively.