以案例式推理法建構類比積體電路在應用端之失效診斷系統

Abstract

知識是企業最重要的資產，更是企業的核心競爭力。然而，知識若缺乏系統化有效的管理，則企業寶貴的資產將在無形中流失。透過有效運用知識管理，可整合、利用組織內部與組織外部的知識，以發揮最大的知識力量。案例式推理(Case-Based Reasoning)是人工智慧的一環，近年來常被使用在知識管理上，此方法是將過去解決問題的經驗儲存成案例庫，當新問題發生時，透過相似度的計算從案例庫中找出相似的案例，做為解決新問題的參考，並依照解決新問題的結果修正找出的的案例，同時將新案例及其解決方法回存至案例庫中，使得案例庫因為持續地學習，進而改善其運作效益。本研究的主要目的就是利用案例式推理的方法建構類比積體電路在應用端的失效診斷系統，做為應用工程師處理新失效問題時參考之用以及規劃員工教育訓練相關課程之用，以縮短應用工程師在處理失效問題的時間以及提升員工教育訓練的效益。本研究最後以一家類比積體電路公司的實例驗證了本研究方法確實具實際效益。

Knowledge is one of the most important assets for an enterprise and also the core competitiveness as well. Knowledge is easy to be lost without an effective and systematic management. An excellent integration for internal and external knowledge could be achieved with suitable knowledge management (KM) technique and to maxize the knowledge power. Case-Based Reasoning (CBR) is one part of Artificial Intelligence (AI). CBR has been used in KM for years. CBR constructs casebase as per the past experience of handling problems in cases. When a new problem comes out, similar cases could be retrieved from the casebase for reference through the similarity calculation. The experience of solving the new case could be also restored in the casease to keep the casebase continuous learning and improving. The main objective of this study is to use CBR to establish a failure-diagnosis system of Analog IC in field application. Furthuremore, the proposed method could be implemented for problem-solving to reduce the time of shooting failure. This system is also good for a new employee at the learning stage. A practical case from a domestic Analog IC company is used to illustrate the effectiveness of the proposed method.