透過您的圖書館登入
IP:54.175.59.242
  • 學位論文

將非常態平均數管制圖設計中成本與監控效率最佳化之研究

A Study of X-bar Control Charts Design with Non-normally Data for Optimizing Cost and Monitoring Efficiency

若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


傳統管制圖經常假設樣本之量測值為常態分配,可是許多實務上的案例卻非如此。在這些實例中常常由於樣本組內之量測值較少,無法應用中央極限定理,使得常態分配的假設並不恰當。 由於較少文獻探討具警告界限之平均數管制圖經濟─統計性設計,因此本研究主要是探討在非常態性資料對於具警告界限之平均數管制圖經濟─統計性設計的影響。選擇經濟─統計性設計模式除了可以保有管制圖預期應有的偵測能力外,同時也能在經濟因素考量下,降低所需付出之成本。本研究在經濟性設計的部份是以Gordon 和Weindling(1975)成本模式,並以生產每件良品之平均成本為績效衡量指標;而統計性設計的部份則是採「平均連串長度」作為統計限制條件。本研究應用基因演算法進行求解,也就是要以基因演算法在滿足所設定的統計限制條件下搜尋管制圖最適參數組合:樣本大小(n)、抽樣間隔(s)、連串長度(r)、警告界限係數(w)與管制界限係數(k),使得生產每件良品的平均成本最小化。 研究結果顯示,固定抽樣成本(CF)與調查並消除可歸屬原因的成本(CA2)變動時,對於單位平均成本並無明顯的影響;而變動抽樣成本(CV)、不良品所花費的成本(CD)、調查錯誤警告的成本(CA1)與製程發生隨機偏移的頻率增加時,單位平均成本明顯遞增。另外,製程發生偏移程度與誤差容許率增加時,單位平均成本則明顯遞減。

關鍵字

無資料

並列摘要


Traditionally, when the issue of designing control chart is discussed, one usually assumes the observations in each sampled subgroup are normally distributed; therefore, the sample mean is also normally distributed. Even if the size of subgroup is large enough, the observations will be distributed normally according to the central limit theorem. However, the assumption may not be acceptable in practice. In this research, an economic-statistical design of X-bar chart with warning limit under Non-normal distributions will be developed using the Burr distribution. In the part of economic design, Gordon and Weindling (1975) cost model is used to minimize average cost per part produced. In the part of statistical design, average run length (ARL) is used as the statistical limiting conditions. The genetic algorithm (GA) is adopted to search for the optimal parameters, i.e. the sampling size (n), the sampling interval (s), the run length (r), the warning limit coefficients (w) and the control limit coefficients (k). By sensitive analysis we can get fixed sampling cost (CF) and cost correcting the assignable cause (CA2) don’t affect average cost per part produced. An increase in variation sampling cost (CV), cost of defective product (CD), cost of searching for assignable cause (CA1) and mean number of shift (θ) leads to increase average cost per part produced. In addition, An increase in shift coefficient (δ) and allowable semi-tolerance leads to increase average cost per part produced.

並列關鍵字

無資料

參考文獻


[2] 包益坤,2003,X-bar與R管制圖的聯合經濟統計模式設計應用在服從非常態分配的品質特性,中原大學工業工程與管理系研究所,碩士論文。
[4] Alexander, S.M. Dillman, M.A. Usher, J.S. and Damodaran, B., 1995, “Economic design of control charts using the Taguchi loss function”, Computers in Industrial Engineering, vol.28, pp.671–679.
[5] Bai, D.S., and Lee, K.T., 1998, “An Economic Design of Variable Sampling Interval X-bar Control Charts.”, International Journal of Production Economics, vol.54, pp. 57–64.
[6] Burr, I.W., 1942, “Cumulative Frequency Distribution.”, Annals of Mathematical Statistics, Vol. 13, pp. 215-232
[7] Celano, G. and Fichera, S., 1999,“Multiobjective Economic Design of an X-bar Control Chart.”, Computers and Industrial Engineering, vol.37, pp. 129-132.

延伸閱讀