透過您的圖書館登入
IP:3.144.161.116
  • 學位論文

量測系統分析的評估研究-TS 16949

On Evaluation of Measurement Systems Analysis – TS 16949

指導教授 : 陳文魁 曹以明

摘要


工業界的量測系統分析(Measurement Systems Analysis, MSA),習於採用表格化的方式來評價量測系統分析,通常由TS 16949國際品質系統五大核心工具之一,MSA指導手冊(2010)的評估準則,進行量測系統的評估,而量測過程的其真正統計模型尚未得知,經常使用由實驗獲得數據,透過變異數分析(Analysis of Variance, ANOVA),來分析量測系統的誤差及探討量測數據之變異數的變化。而古典儀錶的一致性分析時,其使用變異數分析之組內相關係數(Intraclass Correlation Coefficient , ICC)的理論來分析,與現代量測系統分析之量測雙現變異(Gauge Repeatability and Reproducibility,GRR)有所不同,本研究就其關聯性予以探討及推導彼此之間關係。 在產業界的量測行為前提,依MSA指導手冊(2010)的評估準則假設,本文提出量測雙現變異(GRR)模擬流程,並建構其對應的統計模擬程序,透過蒙地卡羅模擬方法(Monte Carlo Simulation Method)的意涵,創新的結合統計概念及Excel的函數表示為 f(x)=ROUND (NORM.INV (RAND () ,Mean ,Standard_dev ) ,2 ),其 Excel 的 RAND 隨機亂數函數的運用,可產生隨機亂數,隨機亂數為0~1之間數值,其值可視為累加累積機率函數的機率值,再使用 NORMINV 反函數,得到其累加累積機率函數的機率值與常態分配位置對應之數值,所得位置對應之數值,再利用轉換為標準常態分配的方法 z=((x-μ))⁄σ,與設定量測系統參數之平均數與標準差,可求得其模擬量測系統的量測實驗值。 由上述之 GRR統計模擬程序,設定九種不同的量測行為情境,其九種情境各模擬1000次,所產生的量測系統之 GRR量測實驗值,使用MSA指導手冊(2010)提供長表格法(Long Form Table)的評估準則與雙因子巢狀式設計模式之變異數分析(Nested Design Model ANOVA)方法,評估量測系統之精度的再現變異(Equipment Variation, EV)及同現變異(Appraiser Variation, AV),即為量測雙現變異(GRR),其兩者方法之間差異和敘述性統計量分析,並藉由%GRR量測能力指標評估,進而衡量該量測系統的整體表現,確保量測系統之良窳。

並列摘要


The industry accustomed to form-based approach to evaluate the MSA, usually one of the five core tools TS 16949 international quality system, MSA Guidebook (2010) of the Long Form Table method assessment criteria, to assess the measurement system, however its true statistical model of the measurement process is not yet known. Data obtained from the experiment, is to analysis error measurement system and discuss changes in the number of variations of the measured data by Analysis of Variance (ANOVA). While the reliability measurement of classical instruments, the group apply ANOVA of the correlation coefficient (Intraclass Correlation Coefficient, ICC) theory to analyze with precision variation (EV, Equipment Variation and AV, Appraiser variation) of the measurement system analysis. It differs from Gauge Repeatability and Reproducibility (GRR). This study is to explore and derive the relations between each other on their relevance. In the measuring behavior of industry, according to MSA Guidebook (2010) assessment criteria assumption, this study proposed the GRR simulation process and the corresponding statistical model simulation program was constructed, through the Monte Carlo simulation method innovative combined with statistical probability concepts and Excel's function expressed as f(x)=ROUND(NORM.INV(RAND(),Mean ,Standard_dev),2). It used Excel's RAND random number function and NORMINV inverse function. It can generate random number by RAND function, the value be 0 to 1. The value can be as the cumulative probability of the cumulative probability function value, and then use the NORMINV inverse function to obtain the value of their cumulative probability function of probability values and normal distribution corresponding to the position of the value. It obtained corresponding to the position of values; then to convert the standard normal distribution z=((x-μ))⁄σ, with setting the mean and standard deviation of the measurement system parameters, GRR simulation of experiments measured value can be obtained. This study used the GRR statistical model simulation program, set nine different measurement situations. In each situation, 1000 times simulations are executed to obtain GRR experimental measured values. These values were analyzed with a Long Form Table method assessment criteria provided by MSA Guidebook (2010) and two-way nested design model ANOVA method, to estimate of the evaluation precision of measurement system of equipment of variation (EV) and appraiser variation (AV), is now called gauge repeatability and reproducibility (GRR). To describe basic statistics, and difference between its two methods, and then evaluate performance in the precision performance indicators of the measurement system is to ensure the measurement systems analysis.

參考文獻


林松茂(2008),ISO/TS 16949(2002年版)品質管理系統稽核常見問題研討會心得分享,品質月刊,44(2),32-38
張值輝(2003),量測系統評價–量測能力指標 %P/T 與 %GRR 之探討,中華民國品質學會第 39屆年會暨第 9屆全國品質管理研討會
楊麗伶(2009),讓MSA發揮最大功效,品質月刊,45(11),54-58
陳文輝(2002b),從ISO/TS 16949 談全球佈局,品質月刊,38(4),91-92
陳文進(2010),企業的經營武器–總經理深入運用 ISO 9001 品質管理體系,品質月刊,46(1),43-46

延伸閱讀