• 學位論文

An empirical investigation of FACT Model: A case of World Class Company

Background Business Excellence Model (BEM) is a management tool used by managers to assess their enterprises and to perform the organizational change in order to purse Business Excellence (BE). However, due to BEMs have been widely implemented in the industries over two decades, a ton of experiences and evidences are accumulated, showing some deficiencies in the implementation of this management tool (ex. Black & Crumley, 1997; Fisher et al., 2001; Hill, 1993; Terziovski & Samson, 1999; Wisner & Eakins, 1994). The deficiencies include excessive paperwork, complex procedures, sophisticated assessment criteria, time consuming and quality bureaucracy, etc. In order to respond these deficiencies, an improved comprehensive BEM was proposed, a three-dimensional Business Excellent Model (BEM) called “FACT”. The acronyms “FACT” indicate the abbreviation of “Framework of Business Excellence”, “Add”, “organizational Culture/Characteristics” and “management Tools/Techniques”. Nevertheless there was just a first study of the proposed model FACT, without any evidence found in firms or real case study conducted in order to demonstrate the FACT dimensions in companies. Purpose This research aims to have an empirical case study to demonstrate how the comprehensive model called FACT is integrated in a world class company not only by using BE framework to be guidance and to make assessment of an organization, but also by choosing the appropriate management tools/techniques and cultivating the right organizational culture/characteristics towards BE. Moreover this research intends to explain how the empirical study of FACT Model in a world class company resolved the argument to the deficiencies identified in the conceptual model. Methodology This research is conducted in two phases. In the first phase, there is a study of the three dimensions of FACT model in order to demonstrate the existence of each element in a world class company and how they are implemented. In the second phase, there is an analysis of the relationship among the three dimensions, in order to explain how the dimensions and its relationship respond to the arguments of the deficiencies. Finding The empirical case study demonstrates the presence of a BEM, Management Techniques/Tools and the right organizational Culture/ Characteristics towards BE in a world class enterprise. The research confirms that the FACT model is not a new business excellence concept, it shows that the presence of three dimensions is integrated in a world class company, and that the dimensions have to be considered as a system, working together to reach BE operations. The other contribution is the response to the argument to the deficiencies of BEM by implementing the FACT model in a world class company. Originality/Value This research is the first empirical case study of the FACT model in a real case in the industry. Nevertheless there was the first study of the proposed conceptual model named FACT; there is no any evidence found in firms or real case study conducted in order to demonstrate the FACT dimensions are integrated in companies and whether the model really has a positive impact in the arguments to the deficiencies.

• 學位論文

以ANP建置製造業導入RFID指標分析之決策機制

在歷經2008年及2009年經濟衰退影響，RFID在2010年恢復狀況良好，預測未來五年期間成長最快的應用項目為製造業、商業服務領域、醫療與生命科學產業、零售商店及物流業。 因生產製造業為台灣主要核心產業，遂引發研究探討製造業導入RFID指標分析之決策機制的議題。 本研究首先，廣泛蒐集RFID相關文獻，及各產業導入RFID的應用案例，藉以了解RFID在各產業導入的考量因素，作為分析探討製造業導入RFID評估因素，佐以ANP分析其評估因素之權重。 歸納整合製造業導入RFID評估分為4大構面及12項評估指標權重及排序分別為: 成本面- RFID硬體成本:權重0.19061;排序1、後續維護成本:權重0.0838;排序6、系統整合成本:權重0.1535;排序3 效率面- 提昇作業效率: 權重0.0861;排序5、生產履歷追溯:權重0.0302;排序10、資訊準確即時共享: 權重0.0317;排序9 企業環境面- IT基礎程度:權重0.0354;排序8、公司高層決心: 權重0.1568;排序2、團對組成與技術能力:權重0.1444;排序4 其它考量因素面- 作業環境與干擾問題: 權重0.0186;排序11、合作夥伴技術能力:權重0.0111;排序12、系統擴充彈性:權重0.0577; 排序7

• 學位論文

Integrated Quality Controlled Monitoring of Inventory Accuracy – combining cycle counting and statistical process control

Inventory accuracy is vital in many industrial environments, particularly large distribution and warehouse environments, especially when implementing computerized systems that relay on accurate inventory. Accurate inventory results from the agreement of the recorded stock keeping unit (SKU) level with the actual SKU level, where inaccuracy affects the performance of the supply operation and realization of benefits form inventory control and procurement. Cycle counting is a technique used to combat inventory inaccuracy, but due to its excessive time and resource requirements this technique fails to ensure inventory accuracy, unaided. Therefore, this research develops an approach to control and measure inventory accuracy promoting the techniques of statistical process control (SPC) that combines resource efficiency and practicality. Specifically, the technique of a c chart is utilized, for the evaluation and ongoing monitoring of inventory management, integrated in implementation by sample selection with only cycle counting methods that are used for measuring accuracy. Also, to aid in the search to find special causes, a method for identifying a change point in the data is presented. The tolerance level setting measure of the proposed method is developed to be utilized in setting the standard for the c chart to provide a useful mechanism for analyzing the performance of the inventory management, coupled with a prescribed inventory accuracy measure. For this research, in an inaccurate inventory the number of adjustments for each SKU is the number of non-conformities. The objective of this research is to develop and analyze a SPC outlook for cycle counting by employing the use of the c chart to examine large populations of SKUs modeled by Poisson counting processes. In the implementation, the focus is on creating a workable integration through sample selection with SPC and cycle counting. In analyzing the proposed approach the use of both analytical as well as simulation results are set forth to demonstrate its effectiveness.

• 學位論文

資源限制專案排程問題財務風險最小化之研究

在實務上，專案的執行也伴隨著專案中各個作業現金的進出（cash in- and out-flows），論文的目的即為降低資源限制專案排程問題（resource-constrained project scheduling problem, RCPSP）於專案執行時的財務風險。研究提出的最大化專案現金可用量問題（project cash availability maximization problem, PCAMP）中，每個作業完成時會得到對應的現金收入並可在專案結束之前被運用，因而，作業的現金－時間值（money-time value）即為作業的現金流入值乘上作業完成時間到專案結束的時間區間值，專案的現金可用量（cash availability）之計算則為所有作業現金－時間值的總和，其提供一保守的現金可用量估計，因在多數情形，如利息，現金會隨著時間的進行而變動。 　　第三章提出三個演算法求解處理單組態的PCAMP模式，分別為GRASP（greedy randomized adaptive search procedure）、MA（memetic algorithm）和VNS（variable neighborhood search）。對於每一個演算法，求解的機制與發展皆由標竿測試題進行分析與比較，而最好的結果將作為此演算法的代表與其他兩者相較。基於最後的結果：GRASP為reactive GRASP結合長期記憶表單（long term memory list）與路徑重連（path relinking）技巧；MA則有兩個，一者為結合GRASP產生初始母體與轉進（restart）母體，另一則使用regret biased sampling產生；VNS的代表則為double VNS （DVNS），其採用一小型的VNS作為每個鄰近解的區域搜尋方式。實驗分析採用以ProGen產生的標竿測試題做為比較分析，其結果顯示結合GRASP的混合MA演算法在各個求解品質之中皆表現最佳，而DVNS則最省計算時間。 　　第四章在於求解多組態的PCAMP，研究提出一兩階段的求解方式，第一階段在於決定各作業的組態指派且滿足專案交期與資源限制，找到最小專案執行成本的排程，此專案執行成本來自各作業執行組態的成本總和，而第二階段則在此組態固定之下，以第三章的模式與演算法求解對應的單組態PCAMP。因而，第四章主要目的即在於求解第一階段的問題。此章提出三個演算法，分別是 (1) BBMA：branch and bound procedure用於決定專案組態指派，而MA則求解某專案組態指派下的單組態問題，並將結果回饋予branch and bound procedure。 (2) DPMA：其包含了兩個步驟，前置步驟（preprocessing）以較短暫的BBMA篩選出個體進入母體，接著以兩個母體進行演化步驟。(3) SA-MA：此方法以多重啟始（multi-start）的方式尋找，每一次開始，每個啟始解，SA皆以CPLEX決定符合不可恢復資源限制的背包問題用以決定啟始的專案組態，如BBMA，SA用於決定組態，而MA求解對應的單組態排程問題。三個演算法的評估採用ProGen與RanGen兩個標竿測試題產生器，對於中小型的題目，以BBMABB求得最佳解與與各演算法相較，而大型問題則與CPLEX取得的下限值做為比較基準，並進一步與一目前知名的演算法BPGA（bi-population genetic algorithm），唯BPGA在原文是用以求解多組態最小化專案完工時間問題。實驗結果顯示BBMA在中小型測試題中表現良好，而SA-MA提供一比較的基準並在大型題目的表現上則優於其他演算法。

• 學位論文

應用六標準差改善製程不良率之研究-以偏光板廠之貼合機改善為例

要在市場上具有競爭優勢，不斷的提高產品品質與可靠度外，另外一個關鍵是降低成本。除了降低採購成本，最好的方式就是良率的提升，還需要利用分析技術找出產品的最佳製程能力，以期望能創造顧客滿意並具有競爭優勢。 本文所探討的是針對偏光板後段貼合機製程之不良項目保護面異物，做製程不良率的改善，利用六標準差中DMAIC方法，針對產品在貼合機製程中導致偏光板有保護面異物不良的要因，找出製程的變異，以六標準差的手法步驟進行探討作持續改善。 用實驗設計（Design of Experiments, DOE）來進行保護面異物改善實驗，找出最佳條件組合。改善前偏光板保護面異物不良DPPM為3992 DPPM，期望改善效益20%為目標，保護面異物不良降為3194 DPPM。 實際驗證的結果得到偏光板保護面異物有減少1049 DPPM，改善達26.2%，且檢驗相關製程中均無產生其它品質上的問題。 最後，改善後偏光板保護面異物不良為2943 DPPM，改善效益預估分析年節省約台幣587萬元生產的內部失敗成本，證明此次個案研究能夠有效改善偏光板貼合機製程產出不良項目保護面異物。

• 學位論文

建構一個PSO-SA的最佳化分類器於不平衡序列資料的分類

資料探勘中非常有名的技術「序列模式探勘」與「序列分類」的技術可以解決許多問題，例如顧客行為變化。然而，傳統的方法對於正確分類出少數樣本的能力不足。實際上，類別資料不平衡的問題經常會發生在日常生活中，例如詐騙行為的偵測、醫療診斷、垃圾信件偵測、產品監控與檢測等，因此本研究將針對不平衡序列資料發展出一個有效的分類方法。在本研究中，根據每個序列所屬的類別，AprioriAll演算法將被用來找出該類別的序列樣式，接著使用pairwise coupling方法將原始的多重類別序列資料拆成組合成許多組二類別的資料，針對每一組二類別的資料，我們會使用所提出的FMCIS方法來建構分類器並命名為FMCIS。每一個分類器會先對一條序列產生兩個相似度的值，接著再運用這兩個值去建構模糊偏好關係所需要的單元。然後使用模糊偏好關係將各個分類器所產生出來的單元值進行整合並加以計算，根據本研究所設定的終止條件將可以產生最後的分類結果。為了增加分類的準確性，一種混和的PSO-SA演算法將會被提出來調整FMCIS裡面序列樣式的權重以及模糊偏好關係裡面類別的權重。結果顯示出本研究提出的分類模型可以有效的解決序列資料不平行的分類問題，但是在模糊偏好關係這個部分，類別的權重並沒有辦法能夠有效地提升整體的分類準確率。

• 學位論文

基於熵理論方法之影像定位處理與應用

影像定位(Image Registration)是一種將兩張或多張於相同場景但不同時間或不同感應器所擷取的影像整合為一張之影像處理方法。一般處理影像定位時，會將兩張影像中相同的灰階值畫素或特徵互相匹配以整合成一張影像。根據這個概念，本論文發展一個以熵理論(Entropy)為基礎且結合重疊區域中灰階差異資訊的影像定位方法，其中灰階差異資訊(intensity difference)則是計算兩張影像中重疊畫素灰階或彩色數值的差距。本影像定位方法是持續更新轉換模型中的參數，使得灰階差異資訊的亂度能最小化以將兩張影像做定位。在效能評估中，本研究將所提出的方法與標準化互相關法(normalized cross-correlation)和標準化共同資訊量(normalized mutual information)藉由多組影像來做測試，並且依據三種方法的穩健程度與量化指標來做衡量。根據實驗結果可得知本研究所提出的影像定位方法，在測試影像中效果優於其他兩種常見的影像定位方法。

• 學位論文

A Model and a Solution Approach for the Berth Allocation Problem

A Model and a Solution Approach for the Berth Allocation Problem Student: Horacio Yamil Lovo Gutierrez Advisor：Dr. Yun-Chia Liang. Department of Industrial Engineering and Management Yuan Ze University Abstract In the last decades, Ports’ operations have been a focal point for global supply chain management and logistics network structures yet they allow the efficient flow of merchandise all over the world. The berth allocation problem (BAP) is closely related with the operational performance of any port. It consists of optimally assigning ships to berthing areas along the quay in a port. A good allocation of ships to berths has a positive impact in terminal’s productivity and in improving customer’s satisfaction levels. Therefore, finding valid formulations which captures the nature of the problem and accounts for the interest of ports operation management is imperative for practitioners and researchers related with this field. This research deals with the discrete version of the berth allocation problem with dynamic arrival times. A new mix integer mathematical formulation is proposed. The model is then compared with other outstanding models from the literature in terms of computational effort. Since the problem is NP-hard, a reduced variable neighborhood search (RVNS) algorithm, employing position based neighborhood structures, is proposed to deal with instances of the problem where exact methods fail to yield a solution in a reasonable amount of time for the decision maker. Furthermore a constructive heuristic is developed in order to improve the convergence speed of RVNS algorithm. Experimental results over instances of the problem, based in information from the ports of Kaoshiung and Gioa Tauro, show that the mathematical formulation proposed not only provides a valid representation of the problem but also an attractive option when compared with other formulations proposed by the literature. On the other hand RVNS offers appealing upper bounds to realistic size instances of the problem while keeping relative low the computation requirements to retrieve a solution.

• 學位論文

以逆向BOM表發展正逆向供應鏈多廠區之訂單分配規畫-應用粒子群演算法

隨著環境問題與有限資源，以及近年來各國相關環保法令相繼提出，故企業開始注重逆向活動與整體封閉迴圈供應鏈之考量。而由於消費型態的改變，導致電子產品的使用週期縮短，廢棄物急速增加，因環境與資源問題，故以生產者的角度，廢棄產品為正向生產活動，帶來了開創新價值的新契機，其廢棄後再使用及再製造可帶來可觀的價值。 而目前生產活動皆以物料清單為基礎，再經由粗略產能規劃與物料需求規劃，進行生產製造；而逆向活動方面，目前並無一套準則來進行逆向活動，故本研究以發展逆向物料清單為主軸，考量封閉迴圈供應鏈中多廠區跨廠之訂單分配規畫，而本研究主要以製造之角度，來進行研究；逆向物料清單主要考量逆向活動中之五種分類方式，分別為回收再利用、回收再製造、回收不可用、有毒廢棄物、一般廢棄物，成本方面考量上述五種逆向活動及正向活動之延遲成本、運輸成本、製造成本、物料成本外，也考量了再利用、再製造與回收不可用所帶來之效益與價值，成為本研究封閉迴圈供應鏈之目標式，並且利用粒子群演算法求解，以求得封閉迴圈供應鏈中總成本最小化為目標。

• 學位論文

應用分析網路程序法於頻率控制元件研發方案之決策分析

現今產業競爭劇烈，企業需仰賴持續研發新產品或改善產品功能，以維持競爭力。本研究以一個生產傳統頻率控制元件廠商為研究個案，使用德爾菲法與網路程序分析法(ANP) 為工具，針對下列三個研發方案對此廠商提供建議：全矽微機電系統振盪器(MEMS OSC)、高階振盪器模組(ADV OSC)、微型化石英晶體振盪器(Mini OSC)。過程諮詢三個部門 - 研發、管理、行銷 - 之主管階層們意見，採用德爾菲法決定六個重要方案評估準則，同時以問卷方式來瞭解他們從這些準則角度來衡量三個構面在達成各個準則之相對重要性。由ANP法所計算之各準則權重，可以提供此廠商研發項目優先順序以及開發具有潛力產品。 研究結論如下：相關構面中以研發構面為最主要之關鍵，管理構面次之，選擇專案時以決定大方向的組織策略為首要考量，其次為金錢預算，再來為技術水準。另外，新技術MEMS OSC為最具優勢之研究方向，其次為Mini OSC。