令 T 表示系統 (system) 的壽命隨機變數,此系統在時間t的可靠度 (reliability) 定義為 ,即在時間t 時系統仍可正常使用的機率。在給定時間t時,可靠度亦可視為一具有先驗分配 (prior distribution) 之隨機變數。由於可靠度以不發生失效之機率作為其定義,是介於0與1的機率值,而Beta分配是定義在 [0,1] 上的機率分配,所以考慮將可靠度視為一具有以Beta分配為先驗分配的隨機變數。 本研究先針對由兩元件串聯所成之系統,給定兩元件在時間t之可靠度分別為 , , 時,且互相獨立時,利用平滑迭代演算法估計系統可靠度 的機率密度函數。其次,再找出使與 的機率密度函數 機率密度函數的 最小的參數 ,並以 作為系統可靠度 之機率密度函數之估計。 此估計方法可以得到機率密度函數的參數估計,也可以推廣至所有串聯、並聯、串並聯、與串聯系統可靠度機率密度函數之參數估計。最後,本研究以Alpha System (Duran & Booker, 1988) 為例,給定各元件可靠度之Beta分配,衍生大小為77的樣本,進行系統可靠度Beta分配之參數估計。
The reliability of a system at time t is defined as , where T is the random variable of the lifetime of the system; that is, the probability that the system can be functioned as designed until time t. For a given time t , reliability can treat as a random variable with a prior distribution. Since reliability is defined as the probability of a system to consistently perform and maintain its function without failure, it ranges from 0 to 1. And the domain of a Beta distribution is also between 0 and 1. The reliability of a system is then considered as a random variable with Beta prior. This research focuses on the beta estimation of the system reliability for a series system with two independent components first. Let the distributions of component reliability be , , ; the system reliability be . The smoothed recursive algorithm is applied to estimate the probability density function of . Then the parameters of beta distribution are chosen to minimize the ; and is treated as the estimation of the system reliability. This method can be expanded to the parameters estimation of system reliability of all the series, parallel, series-parallel, and parallel-series system. Finally, the Alpha System (Duran & Booker, 1988) is adopted as an illustration for this approach. Given the Beta distribution of each component, a sample with size 77 is generated to estimate the Beta parameters of the system reliability.