| 姓名 | 謝昆哲(Kun-Che Hsieh) | 電子郵件信箱 | E-mail 資料不公開 | |
| 畢業系所 | 營建工程系碩士班(Department and Graduate Institute of Constrction Engineering) | |||
| 畢業學位 | 碩士(Master) | 畢業時期 | 95學年第2學期 | |
| 論文名稱(中) | 應用快速混亂基因演算法於計數型雙次抽樣驗收計畫之研究 | |||
| 論文名稱(英) | Applying Fast Messy Genetic Algorithm on the Design of Attribute Double Sampling Plan | |||
| 檔案 |
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。 |
論文使用權限 | 校內外完全公開 | |
| 論文語文/頁數 | 中文/105 | |||
| 摘要(中) |
營建工程品質之檢驗端賴良好的驗收計畫，驗收計畫依其性質可分為計數型與計量型兩大類。計數型抽樣計畫因使用時無須繁雜之統計計算，檢驗時計算簡單，檢驗設備也較為簡易，適用於可採用非破壞檢驗材料的驗收，例如：磁磚、玻璃、燈管及隔間板等。 計數型抽樣驗收(Attribute Sampling Plans, ASP)依其檢驗程序與次數可分為單次與雙次抽樣檢驗形式等，一般而言，雙次抽樣比單次抽樣可節省總驗收成本，因此實務上應用雙次抽樣較單次抽樣多，不過，雙次抽樣驗收計畫設計因需先擬定允收品質水準(Acceptable Quality Level, AQL)與拒收品質水準(Rejectable Quality Level, RQL)，並考慮承包商風險(α Risk)及業主風險(β Risk)，之後求取符合通過操作特性曲線上(AQL, 1-α)和(RQL, β)兩個座標點的允收參數(n1, n2, c1, c2)。此外，允收參數又必須全部是非負值之整數，因此在求解上有其困難度，傳統上曾常用試誤法或啟發式法則來搜尋可能解，然而，使用試誤法或啟發式算法不僅費時且並不保證搜尋的解為最小抽樣樣本數。 本研究乃應用快速混亂基因演算法(Fast Messy Genetic Algorithms, fmGA)，整合α與β風險誤差減至最小與最少的抽樣樣本量等最佳化目標，以求解計數型雙次抽樣驗收計畫(Attribute Double Sampling Plan, ADSP)。本研究以Microsoft Visual Basic 6.0工具語言撰寫應用程式，程式乃提供較彈性的抽樣計畫，使用者可選擇適當的機率分配(超幾何分配、二項分配、卜氏分配)，並輸入業主與承包商雙方於合約內所擬定的AQL和RQL品質要求門檻高低與α和β欲承擔之風險要求，進行目標值的計算，搜尋出最佳適存值之允收參數(n1, n2, c1, c2)解，程式並會將其演算結果繪製成世代收斂圖，方便使用者判定最佳解收斂狀態，以供決策者參考使用。 |
|||
| 摘要(英) | An
effective acceptance sampling plan can control the construction
quality. There are two categories of acceptance sampling plan,
attribute sampling plan (ASP) and variable sampling plan (VSP).
The ASP does not need a complicate statistic calculation and
relatively easy to be used. It is usually applied in the
Nondestructive Testing Evaluation. Depending on the number of samples to be taken from the lot and test procedure, ASP can be classified as two types: single sampling plan and double sampling plan. Generally, the double sampling plan can save more testing cost than single sampling plan. Thus, application of double sampling plan is more popular than single sampling plan in practice. However, the acceptable quality level (AQL), rejectable quality level (RQL), and producer’s risk (α) and consumer’s risk (β) have to be considered in the design of Attribute Double Sampling Plan (ADSP). The combination of the acceptance parameters (n1, n2, c1, c2) have to be found on the operation characteristics (OC) curve and fit the predefined (AQL, 1-α) and (RQL, β). In addition, all of the acceptance parameters should be nonnegative integers. Therefore, it is difficult to find an appropriate solution. Traditionally, the trial-and-error method and the heuristic rules are usually used to find a solution. However, these two methods are time-consuming and plans of minimum sample sizes can not be assured to be reached. In this study, we use Fast Messy Genetic Algorithms (fmGA) as a research method. Objectives of lowering deviation of risk levels (α, β) and minimizing sample size are integrated to find the ADSP. Users can choose a proper probability distribution (hypergeometric distribution, binominal distribution, and Poisson distribution) first. Then, enter the value of AQL, RQL, risk levels (α, β) that both a producer and a customer agree in the contract. Finally, the computer program designed for this study can proceed and find the optimal acceptance parameters (n1, n2, c1, c2). |
|||
| 關鍵字(中) |
|
關鍵字(英) |
|
|
| 口試委員 |
|
|||
| 口試日期 | 2007-07-31 | 繳交日期 | 2007-08-21 | |