運用資料探勘於半導體機台維修管理之知識支援

Abstract

機器設備是半導體產業重要的資產，高精密附加價值的機器設備更需要完整的維護及管理以滿足組織生產營運的需求。設備狀況影響著製造生產系統之產能及產品之良率，所以良好的設備維修保養及管理是企業在競爭日益激烈的市場環境下生存之重要關鍵。在現今高科技產業同業競爭激烈且競相挖角之下，人才流動率普遍相當高，員工一旦離職，其相關的維修知識與能力亦隨即在組織中消逝，知識的再利用效果亦不佳。 資料探勘(Data Mining)是目前在資料分析運用上較常用的方法，目的是將一般資料庫中看似無用的資料(Data)轉換成有用的資訊(Information)，而關聯式法則(association Rule)可挖掘出在關連式資料庫中某些項目間彼此的關係，本研究應用關聯式法則的技術，挖掘了故障碼與維修碼的關連性，將此關聯存成文字檔，再撰寫程式，發展出一套機台維修的故障診斷系統，在維修工程師維修機台時，系統可提供相關的維修建議，一方面可以減少工程師維修故障機台的時間，提高維修效率，增加機台的稼動率，另一方面亦可將相關知識儲存於組織中，降低員工離職對公司所造成的傷害。

Semiconductor equipment is important asset in the semiconductor industry. Semiconductor equipment with high precision added value requires a more complete level of maintenance and management to satisfy organizations’ manufacturing operation needs. The condition of Semiconductor equipment affects the productivity and products’ pass rate of the manufacturing production system. Therefore, sound semiconductor equipment maintenance and management are key factors affecting enterprises’ viability in the increasingly fierce competitive market environment. Nowadays, though aggressive competitions and human resource mining prevailing in the high-tech industry, the job-change rate in general is very high. Once an employee leaves a job position, his/her maintenance knowledge and relevant skills will also soon disappear in the organization, and the reuse effect of the knowledge and skills is usually unsound. Data Mining is a more common approach these days, which is often adopted for data analysis. This can be achieved by employing Association Rule mining to discover the correlation among some items in databases. This research applies the Association Rule mining techniques to discover the correlation between error codes and maintenance codes. By storing the correlation as a word document, this work can then develop a fault diagnosis system for semiconductor equipment maintenance. This system can provide relevant repair suggestions to engineers performing repair functions. The proposed system can reduce the engineer’s repair time on the faulty machine, to increase maintenance efficiency and machine utilization rate; the system can also store the relevant knowledge in the organization, thus minimize potential damages to the company caused by employees’ resignation