整合多準則決策與失效模式於液晶顯示器之製程研究

Abstract

失效模式分析(FMEA: failure mode and effects analysis)已被廣泛地應用在許多領域，包括航空、軍事、汽車、電子、機械與半導體產業等，用來辨識及消除潛在的劣化因子或風險因素等。一般來說，傳統的失效模式會依據所謂的風險優先指標(RPN: risk priority number)對失誤事件(failure events)進行排序，而風險優先指標即由三項因素的簡單乘積來計算，包括發生率(occurrence)、可測度(detection)與嚴重性(severity)。很明顯地，傳統失效模式分析並未考慮三項風險因素的相對重要性，致使區辨不同風險組合所造成的相同風險優先指標之失誤事件變得非常困難；縱使其背後所隱藏的物理意義可能完全不同，而且追蹤失誤事件的根本成因(root cause))在實務上也變得極不可行。因此，本研究提出一個統合的多準則決策(MCDM: multi-criteria decision making)模型，以模糊尺度的語意量表；將層級分析法(AHP: analytical hierarchy process)、灰關聯法(GRA: grey relational analysis)、與決策實驗分析法(DEMATEL: decision making and trial laboratory)彙整以克服傳統失效模式的困境，同時以液晶顯示器的製程實證資料來印證所提架構之有效性。

Failure mode and effects analysis (FMEA) has been widely applied to many domains (i.e. aerospace, military, automobile, electronic, mechanical, and semiconductor industries) for identifying and/or eliminating potential deteriorated factors, risks, and problems. Generally, the conventional FMEA prioritizes specific failure events based on a so-called risk priority number (RPN), which is a multiplicative product of the three risk factors: occurrence (O), detection (D), and severity (S). Obviously, the conventional FMEA is deficient in considering the relative importance among occurrence, detection and severity and incapable to distinguish different combinations of three risk factors resulting in the same RPN value. Hence, linking specific failure events to corresponding causal factors becomes very difficult and infeasible in practice. As a result, a fuzzy MCDM (multi-criteria decision making) based approach that incorporates AHP (analytical hierarchy analysis), GRA (grey relational analysis), and DEMATEL (decision making trial and evaluation laboratory) is presented to overcome the above-mentioned shortcomings. An industrial example regarding the fabrication process for TFT-LCD (thin film transistor-liquid crystal display) is demonstrated to validate the proposed approach.