頁籤選單縮合
題 名 | 品管手法中層別法之最適抽樣數之決定=Determined the Optimal Sampling Numbers of the Stratification Method in the Quality Control Tools |
---|---|
作 者 | 盧昆宏; 邱千惠; | 書刊名 | 管理與系統 |
卷 期 | 7:3 2000.07[民89.07] |
頁 次 | 頁443-458 |
分類號 | 494.56 |
關鍵詞 | 品管手法; 層別法; 分層權重; 特定比例之層別; 多變數層別; Quality control tools; Stratum; Stratification; Stratum weight; Stratified for proportion; Multiple variables of stratification; |
語 文 | 中文(Chinese) |
中文摘要 | 品管手法中之層別法係依產品之原料,或生產之機械,或操作人員或操作方法…等因素,分別來蒐集數據進而找出其間的差異,並針對差異再加以改善的方法。簡言之,為區別各種不同原因對於結果之影響,而以個別原因為主體,分別作統計分析的方法謂之層別法。該品管方法在1996年以前僅被用於教導現場實務人員對製程因素之分類而已,並沒有去探討分層後蒐集數據時之檢驗成本及各層的最適抽驗個數應該為多少才可以達到產品預期之出廠變異數,及滿足最低的抽驗成本。直到Chang and Lu(l996)引入一線性成本函數,而建構出一套最適之層別程序,在該程序中,分別推導出當平均數之變異數固定時,最適之抽驗個數;以及當總檢驗成本固定時之最適抽驗個數。然而,時至今曰仍沒有文獻探討,當吾人利用層別法來進行現場產品品質追蹤及改善時,需同時考慮數個變數時之最適抽樣數及相關統計量等問題;層別後實際抽驗與最適抽樣之間誤差的探討;以及母體具特定分層比例時,其各分層的最適抽樣數應為何等均沒有被討論過。因此,在本文將透過統計理論之推導,計算出層別後實際抽驗與最適抽樣之間的誤差及其均力誤(MSE),按著,探討母體中各分層具特定比例時之最適抽樣數,最後導出多變數層別時之最適抽樣數之公式,並探討其性質,俾供品質追蹤與改善之用。 |
英文摘要 | The quality of products is often influenced by the following factors: materials, components, equipment, operators, operating methods, climate, humidity, and so on. The quality can be improved only by digging out the real cause. An approach called stratification can be used to describe the systematic subdivision of population or process. Proper stratification shows the root causes of the problem and leads the way for the establishment of proper countermeasures. Chang & Lu (1996) proposed a practical stratification procedure to infer the sampling number of each stratum by keeping the variance of the estimated mean constant or infer the sampling number of each stratum by keeping the sampling cost. Nevertheless there are still no relative literatures to illustrate the following things: (1)the problem of optimum sampling number with more than one quality characteristic in stratification, (2)the problem of stratified sampling for proportions, and (3)the effects of deviations from the optimum allocation. Consequently, in this research, authors synthesize the concept of quality management, sampling theory and statistics theory to discuss the effect of errors between the stratum optimum sampling sizes with the stratum random sample sizes, the stratified optimum sampling numbers for proportions, and the problem of the stratified optimum sampling numbers with multiple variables. A linear cost function is derived to study the above aspects of the stratified optimum sampling sizes of each stratum. |
本系統中英文摘要資訊取自各篇刊載內容。