應用 DBSCAN 演算法以提升生產線 UPH 驗證準確率

NCUIR > School of Management at National Central University > Executive Master of Industrial Management > Electronic Thesis & Dissertation > Item 987654321/94598

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/94598

Title:	應用 DBSCAN 演算法以提升生產線 UPH 驗證準確率
Authors:	徐立宇;HSU, LI YU
Contributors:	工業管理研究所在職專班
Keywords:	演算法;聚類演算法;分群;四分位數;DBSCAN;UPH;K-means;OPTICS;PCB;Tukey
Date:	2024-07-25
Issue Date:	2024-10-09 15:18:16 (UTC+8)
Publisher:	國立中央大學
Abstract:	本研究探討了應用 DBSCAN 演算法來提升生產線 UPH（每小時產量）驗證準確率的有效性。電路板（PCB）作為電子產品中的核心部件，其需求急劇增長，對於製造公司來說，準確預測和監控生產線產能至關重要。傳統的數據分析方法，例如 Tukey 方法，雖然在處理離群值方面有一定成效，但是在面對數據有多峰分佈和非對稱數據時，往往會導致計算結果不夠精確。因此，本研究基於多項數據演算法的評估下，提出了一種基於 DBSCAN（密度聚類演算法）的數據分析方法，目的在於改進現有的數據驗證方法，提高生產線產能的計算準確性。在研究過程中，我們首先介紹了 PCB 生產線的運作模式及相關產能計算的基本概念，接著回顧了現有的數據分析方法及其局限性，並詳細比較了多種分群演算法，最終選擇 DBSCAN 演算法作為研究主軸。通過對實際生產數據的測試和評估，新方法能夠更準確地識別和排除離群值，從而提高數據驗證的準確性和可靠性。研究結果顯示，與傳統方法相比，應用 DBSCAN 演算法能顯著提高 UPH 的驗證準確性，對於製造公司的生產計劃和決策具有重要的指標意義。 ;This study explores the effectiveness of applying the DBSCAN algorithm to improve the accuracy of UPH (units per hour) verification in production lines. As printed circuit boards (PCBs) serve as the core components in electronic products, their demand has surged dramatically. For manufacturing companies, accurately predicting and monitoring production line capacity is crucial. Traditional data analysis methods, such as the Tukey method, while somewhat effective in handling outliers, often lead to imprecise results when dealing with multi-modal and asymmetrical data distributions. Therefore, this study proposes a data analysis method based on the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, evaluated against multiple data algorithms, to improve existing data verification methods and enhance the accuracy of production line capacity calculations. During the research process, we first introduced the operational model of PCB production lines and the basic concepts of related capacity calculations. Then, we reviewed the existing data analysis methods and their limitations, and thoroughly compared various clustering algorithms, ultimately selecting the DBSCAN algorithm as the main focus of the study. Through testing and evaluating actual production data, the new method can more accurately identify and exclude outliers, thereby improving the accuracy and reliability of data verification. The research results indicate that, compared to traditional methods, applying the DBSCAN algorithm significantly enhances the accuracy of UPH verification, providing crucial insights for manufacturing companies′ production planning and decision-making.
Appears in Collections:	[Executive Master of Industrial Management] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	38	View/Open

社群 sharing

Loading...