可修復系統新方法之發展與預測模式之建立

Abstract

根據產品毒命曲線的特性，任何產品的故障率都會隨時間的增加而變大，而使系統的可靠度 (reliability) 漸漸退化。若要減少系統可靠度的退化，則必須重新設計系統，以消除系統的特定失效模式。然而，此舉必須對系統作實際的重新測試，因此相當花成本和人力。 由於上述的缺失，本論文乃針對一可修復系統 (repairable system)，應用Nelson於1988年所發展的非參數圖形法 (nonparametric graphical method) 及揩數平滑法 (exponential smoothing method) 的技巧，以消除原有系統的特定失效模式，如此只需直接校訂系統的修復數據即可，不必對系統作重新的測試，且可改善原設計系統的可靠度，因此可節省成本和人力。 除此之外，根據本論文的新方法所校訂之修復數據，利用時間數列分析 (time series analysis) 的方法，可合理地預測系統未來的平均修復數，以瞭解系統未來修復率之趨勢。最後，本論文利用汽車修復數據的實際資料，驗證了所構建的新方法和預測模式之有效性及可行性。

Through the characteristic of the life curve of products, the failure rate of products increases with the increase of the time. Therefore, the reliability of the system will degrate. If we want to reduce the degration of the reliability, we have to redesign the system to eliminate the certain failure mode. However, this requires to retest the system and it is very costly and time-consuming. Because the disadvantage mentioned above, the objective of this thesis is to make use of the nonparametric graphical method, developed by Nelson in 1988, and the technique of exponential smoothing method to eliminate certain failure mode of the original system. In this way, without retesting the system, we may devise the repair data of the system directly. Therefore, it can not only improve the reliability of the original system, but also save some cost and time. Also, based on the repair data of the new approach developed in this thesis, we may apply the method of time series analysis to predict the mean repair numbers in the future time period to realize the trend of the repair rate. Finally, a real-world repair data of a repairable system from the automobile industry is used to verify the effectiveness of the approach and the predicted model developed in this thesis.