目前的抽樣檢驗計劃,只針對單一組件組合系統且皆以傳統統計的理論根 據為基礎所發展出來的;然而,利用貝氏統計觀點與決策理論的基礎來探 討進貨組件之抽樣檢驗,將成本效益和事前機率分配予以加入,建立使抽 樣期望成本達最小的目標,更臻具體。本論文針對多組件可靠度系統設計 貝氏抽樣檢驗計劃。此檢驗計劃為屬性抽樣(Attribute Sampling)及修正 檢驗(Rectifying Inspection Procedures)。 系統組合組件是良品或不 良品假設為可交換性(Exchangeable)隨機變數。我們對組合系統為由多相 同組件所組合及兩個不相同組件所組合情形分別探討;並提出各自由模式 分析而導出的演算法則。
The procedure presents two new rectifying inspection and attribute sampling plans for multi-component products. The states of items in the lot are assumed exchangeable. Not like common sampling plans which use classical statistical decision theory, our sampling plans are developed based on Bayesian decision theory of which many people think is more sensible right now. Barlow and Xiang(1986,1987)developed two Bayesian sampling plans for the case that a product contains a component. We advance their research. We study the case that a product containts several identical components and the case that a product contains two different components. The algorithms for these two cases are developed based on model analysis.