抽樣檢驗是品質管制的重要一環，基於成本、時間或現實需要的考量，廠商無法對所有產品逐一檢測，因此會在整批產品中隨機抽取部份樣本進行檢驗，並將檢驗結果與檢驗標準做比較，藉此判斷該批產品應該全數允收或全數拒收，以維持出廠產品的品質。

傳統抽樣檢驗法的檢驗條件均假設為明確的數值，例如︰產品不合格率、生產者風險、消費者風險…等。但在實際的情況下，由於製程容易隨時間發生變化，因此無法得知不合格率的精確數值，僅能知道其大概的數值。而生產者風險及消費者風險，應該由生產者及消費者決定其要求的標準，一般而言皆為明確數值。故本研究基於上述假設，將模糊理論運用於單次抽樣檢驗法，制定通過兩點之模糊抽樣檢驗法，以及道奇與洛敏拒收水準型之模糊抽樣檢驗法。最後將本研究之方法與傳統及其他模糊抽樣檢驗方法進行分析比較，結果顯示通過兩點糢糊抽樣檢驗法較傳統的JIS　Z9002抽樣計畫表有較穩定的生產者風險與消費者風險；而採用重心解模糊觀點的道奇與洛敏拒收水準型之模糊抽樣檢驗法可避免delta-截集方法中不易選取delta的困擾。

Sampling inspection is an important subject in quality controls. Consider the restrictions of cost, time or reality problems, the manufacturer is unable to inspect all products, so they will inspect samples randomly, and compare with test criterion to judge the lot of products should be accepted or rejected. By this process, we can make sure the quality of outgoing productions.

The observations of traditional sampling inspection are supposed precise number, for instances, the fraction defective p, producer’s risk, and consumer’s risk...and so on. However, it is difficult to keep a process in control with a fixed process level for a long period of time, so we get the vague number p instead of precise one. However, the producer’s risk and consumer’s risk should be decided by producer and consumer, so we consider they are both precise numbers. Based on above assumptions, the fuzzy theory is applied in single sampling inspection plans in this thesis, including the sampling plan of given strength and Dodge and Roming LPTD system.

Also, we compared the proposed methods with traditional and other fuzzy sampling inspection methods. The result shows that the proposed sampling plans of given strength have steadier producer's risks and consumer's risks than JIS Z9002, and the proposed Dodge and Roming LPTD system sampling plan by using the center of area method can avoid the problem on the choice of delta values when using delta-cut method.

[1] Bolshev, L.N., Gladkov, B.V., and Scheglove, M.V. (1961), “Tables for the cal-culations of R- and Z-distributions.” Theory of Probability and its Applications, 6, 410-419.

[2] Chakraborty, T.K. (1988), “A single sampling attribute plan of given strength based on fuzzy goal programming,” Opsearch, 25, 259-271.

[3] Chakraborty, T.K. (1992), “A class of single sampling plans based on fuzzy op-timisation,” Opsearch, 29, 11-20.

[4] Chakraborty, T.K. (1992), “Single sampling LPTD plans as a stochastic pro-gramming problem,” Sankhy , 54B, 242-248.

[5] Chakraborty, T.K. (1994), “A class of single sampling inspection plans based on possibilistic programming problem,” Fuzzy Sets and Systems, 63, 35-43.

[6] Fujino, Y., and Okuno, T. (1981), “On single sampling plans by attributes with given producer’s and consumer’s risk,” Journal of the Japanese Society for Quality Control, 11, 169-177.

[7] Hald, A. (1981), Statistical Theory of Sampling Inspection by Attributes, London: Academic Press.

[8] Kanagawa, A., and Ohta, H. (1990), “A design for single sampling attribute plan based on fuzzy sets theory,” Fuzzy Sets and Systems, 37, 173-181.

[9] Klir, G., and Yuan, B. (2005), Fuzzy Sets and Fuzzy Logic: Theory and Applica-tions, Taiwan: Pearson Education.

[10] Ohta, H., and Ichinashi, H. (1988), “Determination of single-sampling-attribute plans based on membership functions,” Internat. J. Production Res., 9, 1477-1485.

[11] Wang, R.C. and Chen, C.H. (1997), “The Dodge Roming double sampling plans based on fuzzy optimization,” Internat. J. Quality Science, 2, 52-62.

[12] Chen, H. and Wang R. (1999)，Minimum AFI for csp-1 plan based on fuzzy op-timization，模糊系統學刊，第5卷, 63-68頁。

[13] 潘浙楠、李文瑞(2003)，品質管理，華泰文化事業股份有限公司。

[14] 潘浙楠、周芳儀(2005)，模糊歸屬函數在修正計數值抽樣檢驗計劃之應用，品質學報，第12卷，89-100頁。