Please use this identifier to cite or link to this item: `http://hdl.handle.net/11455/17527`
 標題: 使用大量遮蔽系統存活資料來漸近估計個體可靠度Asymptotic Estimation of Component Reliability Using Large Number of Masked System Life Data 作者: 鄭百志 關鍵字: Exponential distribution;指數分配;masked data;maximum likelihood estimation;mixture;unsupervised learning;遮蔽資料;最大概似估計量;混合;沒有監督的學習 出版社: 應用數學系 摘要: 摘要 由多個個體系統的存活資料可估計這個系統中的每個個體的可靠度。由於成本和診斷上的限制，個體的可靠度可能無法獲得；或者無法獲知運作中系統個體的可靠度。 在本篇論文中，我們考慮串聯系統；並且假設每個個體的存活時間遵守指數分配。傳統找尋可靠度的方法，是處理複雜多參數最佳化問題的最大概似估計法。我們在本篇論文中，提供一個簡單概似函數和演算法，使用大量遮蔽系統存活資料來估計每個個體的未知故障率(可靠度)。 最後的數值例子中，使用大量的遮蔽系統存活資料，結果證明每個個體之未知可靠度的估計值和參數值是接近的。Abstract Life data from multi-component systems (or masked system life data) are used to estimate the reliability of each component in a system. Due to cost and diagnostic constraints, the component reliability may not be known or it is impossible to know the component reliability in the systems such as living organisms. In this paper, we consider a system where if one of components fails, the whole system fails. Suppose that the lifetimes of components have exponential distributions. The traditional approach to finding such reliability is the method of maximum likelihood (ML), which is a complicated multi-parameter optimization problem. We propose a simple likelihood function and a simple algorithm to find the unknown failure rate (reliability) of each component using a large number of masked system life data. The results of a Monte Carlo study are presented to demonstrate the convergence of estimates to the unknown reliability of each component using large masked system life test data. URI: http://hdl.handle.net/11455/17527 Appears in Collections: 應用數學系所