透過您的圖書館登入
IP:3.145.47.253
  • 學位論文

以蒙地卡羅模擬評估型 I 設限預測模型並推估保固風險-以DOM產品為例

Utilizing Monte Carlo Simulation on evaluating the Type I Censored mode and warranty risk assessment - Disk on Module for example

指導教授 : 黃乾怡
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


由於近年來新產品研發速度越來越快速,時間成為主宰企業成敗的關鍵因素。除了在短時間內推出良好的產品,在產品的品質上也因為市場越來越競爭,對於產品品質與可靠度的要求越來越嚴格。一般企業在推出新產品會對產品進行可靠度測試與壽命預估,以確保產品品質與產品保固時間。 本研究以微型硬碟記憶卡(Disk on Module, DOM)產品為研究的對象,資料來源為企業售後服務端所搜集的客退資料(Return Materials Authorization, RMA),分析該產品失效曲線,並在保固期間內進行預測。研究內容分為三個部份,第一部分以2004年1月至12月各月份中,搜集各月份下同批出廠產品的失效數據(可觀察資料),也就是搜集12筆樣本,在各筆樣本中的RMA進行遺失值估計,將以估計得到的插補值替代,避免後續分析造成統計偏誤。當可觀察資料與遺失資料合併後成為兩年內完整的失效資料,再分別對12筆樣本以兩年內完整的失效資料找出產品失效所服從的機率分配。 第二部份取其中較能代表失效分佈的分配參數組合,以蒙地卡羅模擬出不同參數組合下的失效狀況,以模擬值作為真值,比較不同型I設限預測模型的預測準確度,藉由條件覆蓋機率與相對誤差來對模型間的預測能力進行比較探討,以找出能夠適當描述產品DOM的設限預測模型。 第三部份探討目前仍在保固期限內的產品,以2006年1月出廠產品為例,先搜集從出廠後的兩年失效數據,估計遺失值以插補值取代,最後以兩年內完整的失效資料由合適的型I設限模型進行預測保固期間內失效個數上界,並提供相關可靠度指標以協助合作廠商在對產品制定保固期限能有更精確的參考依據。

並列摘要


This research has analyzed and focused on the Disk on Module (DOM) and the source of origin is based on the “Return Materials Authorization (RMA)” which had been collected by after-sales service program of the enterprise. Organizing the missing data estimation by adopting the RMA, and then applying the estimate values imputation for missing data in order to avoid the bias which caused from the further analysis! Developing the Monte Carlo Simulation for determining the failure times which are independent observations from Weibull distriburtion and failure distributed under the different parameters. Furthermore, using the simulation value as the real value and comparing the prediction accuracy of different type I censored modes. Next, utilizing the conditional coverage probability (CP) and relative error (RE) to compare and contrast the prediction capability between the modes in order to discover the mainly appropriate type I censored mode for product “DOM”. Lastly, selecting the most appropriate type I censored and processing the prediction test for failure quantity upper bound under the product warranty term and also positively supply the reliability index as the reference in order to assist the enterprise to organize and establish the proper warranty term for their latest products.

參考文獻


[24] Balakrishnan, N., and Aggarwala, R., Progressive Censoring-Theory, Methods, and Applications, Boston: Birkhauser, 2000.
[1] Hussain, A. Z. M. O. and Murthy, D. N. P., "Warranty and optimal reliability improvement through product development," Mathematical and Computer Modelling, vol. 38, 2003, pp. 1211-1217.
[3] Rubin, D. B., "Inference and missing data," Biometrika, vol. 63,no. 3, 1976, pp. 581-592.
[4 ]Musil, C. M., Warner, C. B., Yobas, P. K., and Jones, S. L., "A comparison of imputation techniques for handling missing data," Western Journal of Nursing Research, vol. 24, no. 7, 2002, pp. 815-825.
[5] Little, R. J. A., and Rubin, D. B., Statistical analysis with missing data, New York: John Wiley, 1987.

被引用紀錄


王致達(2012)。建構產品活化能估算及可靠度驗證系統-以DOM產品為例〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://doi.org/10.6841/NTUT.2012.00088
劉玉萍(2017)。偏光片可靠度加速壽命試驗研究〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840/cycu201700728
李淑華(2012)。底膠製程優化與良率模型建立-以智慧型手機為例〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0006-1108201212110200
蔡佩育(2013)。利用客退資料作雲端產品壽命分析-以KVM 切換器為例〔碩士論文,國立臺北科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0006-0307201316470600

延伸閱讀