A research of performance data collection’s planning and analyzing on Reliability of product and medium match up

[[abstract]]產品依其失效的特性，其失效率函數呈一浴缸型曲線形狀，該浴缸型曲線則隨不同之失效數據而異，通常失效率函數可由產品壽命測試數據獲得，且產品失效分佈可分為早夭期、平穩期及老化期，然而在產品可靠度退化模式中，產品之失效分佈函數將隨著不同之產品類型而有不同的狀態，由過去之文獻中常被應用之壽命分佈，如常態分佈、對數常態分佈、韋伯分佈、指數分佈…等，這些壽命之分佈作為代表各種產品在使用過程中的失效行為，以了解產品在使用過程中其可靠度之變化。 針對評估產品可靠度的統計分析之前，必須先選擇適當的壽命分配當作往後統計分析的基礎。然如何選擇合適的失效分佈來當作推論壽命試驗失效數據的分配，這是一重要的課題，因為錯誤的選擇將影響推論的結果。本研究將應用適合度檢定之技術如K-S檢定方法來判斷產品之壽命實驗數據所建立的模式何種為最適合之失效分配，同時觀察其耗用介質之可靠度退化行為以及推測介質之最佳置換時機。[[abstract]]According to the failure characteristic of product, the hazard function can be described by bathtub-shaped curve. Its shape characteristics depend on the failure data. Generally, the hazard function of product can be gotten by the data of product life test and it can be classified into three regions, early life region, constant failure rate region and wear out region. But, in the product reliability deterioration model, the different products have different life distributions. Most of the failure distribution in the literatures are applied such as Normal Distribution, Lognormal Distribution, Weibull Distribution, Exponential Distribution etc. These distributions can be represented the failure process of the variety products and understood the change of the reliability during the applied duration. At first, we shall select appropriate the failure distribution as the base of statistic analysis before we launch to evaluate the reliability of the product. How to select an suitable failure distribution to be the foundation of the inference of the life distribution will be an important issue. Because of the wrong choice will cause the error results. The paper will be applied the goodness-of-fit test such as K-S method to evaluate the most suitable failure distribution based on the life test data to build up model. And we can observe the deterioration of the reliability with the consumable medium and estimate the optimum replaced timing of the medium