| dc.contributor.advisor | 羅明琇<br>郁方 | zh_TW |
| dc.contributor.advisor | Lo, Ming-Shiow<br>Yu, Fang | en_US |
| dc.contributor.author (Authors) | 林書琪 | zh_TW |
| dc.contributor.author (Authors) | Lin, Shu-Chi | en_US |
| dc.creator (作者) | 林書琪 | zh_TW |
| dc.creator (作者) | Lin, Shu-Chi | en_US |
| dc.date (日期) | 2022 | en_US |
| dc.date.accessioned | 1-Aug-2022 19:00:55 (UTC+8) | - |
| dc.date.available | 1-Aug-2022 19:00:55 (UTC+8) | - |
| dc.date.issued (上傳時間) | 1-Aug-2022 19:00:55 (UTC+8) | - |
| dc.identifier (Other Identifiers) | G0109363098 | en_US |
| dc.identifier.uri (URI) | http://nccur.lib.nccu.edu.tw/handle/140.119/141398 | - |
| dc.description (描述) | 碩士 | zh_TW |
| dc.description (描述) | 國立政治大學 | zh_TW |
| dc.description (描述) | 企業管理研究所(MBA學位學程) | zh_TW |
| dc.description (描述) | 109363098 | zh_TW |
| dc.description.abstract (摘要) | 全球製造業朝向智慧轉型發展，伴隨物聯網架構、大數據雲端運算系統、人機協同系統、智慧設備等技術成熟，創造新型態的製造環境。但現行製造產業大多僅透過自動化設備進行生產，雖能提升生產力，卻無法提高品質管理效率，仍須耗費人力成本及時間在品質檢測上，且人工抽檢並無法全面且即時的掌握品質狀況，如此一來既會產生不良品成本，更可能使整體品質水準下降，使公司承受聲譽受損的風險。本研究透過安裝感測器搜集廠內製造環境大數據，以數據模型作為工具輔助企業進行流程再造，消除製造流程中的浪費使之達到精實生產的目標。以雙向深度長短期記憶模型（Bi-directional LSTM Model）作為基礎，加上兩個邏輯規則及其比重 (α 及 β )的模型有最好的預測效果，整體精準度達97.87%。代表本模型能作為科技媒介在流程再造中發揮效果，有效的優化並取代傳統生產流程，不但能夠降低瑕疵品風險，更能降低品質管理成本及其他成本、有效減少浪費，落實精實生產（lean production）的核心目標。此外品質肇因分析模型以可解釋人工智慧架構（Explainable AI）做為基礎，發現轉速對於整體模型的貢獻度最高而頻率及狀態佔比極低，代表轉速很可能是影響品質優劣的原因，後續製程優化亦可從轉速（Speed）作為切入點進行分析，將能有效提高製程優化效率而縮短製程優化研發時間及成本。 | zh_TW |
| dc.description.abstract (摘要) | The manufacturing industry around the world is developing with digital transformation. With the technologies such as the Internet of Things, big data and cloud computing system, human-robot collaboration system and smart device, it create a new type of manufacturing environment. However, most manufacturing companies only use automated equipment for production but few use other technologies. It can improve productivity, but cannot improve the efficiency of quality management. It still takes labor costs and time to do quality inspection. In addition, sampling inspection by people cannot understand the quality status comprehensively and instantly. It make the quality level lower and incur the cost of defective products, more likely to expose the company to the risk of reputational damage.Therefore, this research will aim to reduce process waste and assist enterprises in process reengineering through data modeling. By installing sensors to collect the big data of the manufacturing environment in the factory, Combination Rule Bidirectional LSTM Model is used to build a quality prediction model. The prediction accuracy of the Model achieves 97.87%. It means that this model can be used as a technological medium in process reengineering to effectively optimize or replace traditional production processes. It can not only reduce the risk of defective products, but also reduce the cost of quality management. Companies can easily reduce waste and implement the core goal of lean production.In addition, the quality cause analysis model built by Explainable AI can provide analysis suggestions for process optimizers. It can effectively improve the efficiency of process optimization and shorten the development time and the cost. For example, it can be found from the model that the speed has the highest contribution to the overall model, and the frequency and state proportions are extremely low, which means that the speed is likely to be the reason that affects the quality. Process optimization can be analyzed from the speed as an entry point. | en_US |
| dc.description.tableofcontents | 第一章 緒論 8第一節 研究背景與動機 8第二節 研究目的 11第三節 研究流程 12第二章 文獻回顧 14第一節 精實生產 14第二節 以人工智慧輔助精實生產：精實製造4.0 18第三節 品質管理的新概念：品質預測 20第四節 應用深度學習方法於品質管理 23第三章 以流程再造實現精實生產 27第一節 生產流程再造 27第二節 品質管理模型 30第三節 透過精實生產減少浪費 46第四章 結論與建議 48第一節 研究結果 48第二節 研究貢獻 49第三節 研究限制與改善建議 50第五章 參考文獻 52表目錄表2- 1減少七大浪費 14表2- 2精實生產的五大原則 16表2- 3精實生產的步驟和原理 16表3- 1 資料集分佈情況 31表3- 2「產品規格全檢及良率預測」輸入值（Input）資料清單 32表3- 3「產品規格全檢及良率預測」輸入值（Input）資料清單(續) 33表3- 4不同 α及 β 時之模型準確度 37表3- 5 13個別規格之散佈圖及混淆矩陣 39表3- 6 13個別規格之散佈圖及混淆矩陣（續） 40表3- 7 13個別規格之散佈圖及混淆矩陣（續） 41表3- 8 13個別規格之散佈圖及混淆矩陣（續） 42表3- 9不同預測項目的SHAP value 43表3- 10不同期別各input對模型的SHAP value 44圖目錄圖1- 1研究流程圖 13圖2- 1瑕疵產生及檢測的時間序 20圖2- 2產品中可能出現瑕疵的樹階層結構 22圖2- 3Deep LSTM Model架構圖 25圖2- 5 Bi-directional LSTM Model 模型架構圖 26圖3- 1企業流程再造前後比較圖 29圖3- 2模型建置流程圖 30圖3- 3產品規格全檢及良率預測之x變數以連續序列生成子資料集 34圖3- 4產品規格全檢及良率預測模型設計結構圖 35圖3- 5當不同的 α 及不同的 β 之準確度 38圖3- 6當α=0.1, β=0.1時之損失函數 38圖3- 7當α=0.1, β=0.1時之混肴矩陣 39圖3- 8不同期別對整體預測模型SHAP value 45圖3- 9企業流程再造減少的浪費 46圖3- 10模型建置流程圖 47 | zh_TW |
| dc.format.extent | 2919194 bytes | - |
| dc.format.mimetype | application/pdf | - |
| dc.source.uri (資料來源) | http://thesis.lib.nccu.edu.tw/record/#G0109363098 | en_US |
| dc.subject (關鍵詞) | 品質管理 | zh_TW |
| dc.subject (關鍵詞) | 數位轉型 | zh_TW |
| dc.subject (關鍵詞) | 智慧製造 | zh_TW |
| dc.subject (關鍵詞) | 流程再造 | zh_TW |
| dc.subject (關鍵詞) | 大數據 | zh_TW |
| dc.subject (關鍵詞) | 機器學習 | zh_TW |
| dc.subject (關鍵詞) | Quality Management | en_US |
| dc.subject (關鍵詞) | Digital Transformation | en_US |
| dc.subject (關鍵詞) | Smart Manufacturing | en_US |
| dc.subject (關鍵詞) | Process Reengineering | en_US |
| dc.subject (關鍵詞) | Big Data | en_US |
| dc.subject (關鍵詞) | Machine Learning | en_US |
| dc.title (題名) | 透過大數據建模建立智慧製造之品質管理解決方案 | zh_TW |
| dc.title (題名) | Using Machine Learning in building a Quality Management System for Intelligent Manufacturing | en_US |
| dc.type (資料類型) | thesis | en_US |
| dc.relation.reference (參考文獻) | 一、 英文參考資料Achinthya D. Perera, Nihal P. Jayamaha, Nigel P. Grigg, Mark Tunnicliffe, Amardeep Singh (2021). The Application of Machine Learning to Consolidate Critical Success Factors in Lean Six Sigma, IEEE Access, vol. 9: 112411-112424András Pfeiffer, Dávid Gyulai, Gábor Nick, Viola Gallina, Wilfried Sihn, and László Monostori (2018). Lead time prediction in a flow-shop environment with analytical and machine learning approaches. IFAC-PapersOnLine, vol 51, Issue 11: 1029-1034Beata Mrugalska*, Magdalena K. Wyrwicka (2017). Towards Lean Production inIndustry 4.0, Procedia Engineering Vol.182: 466-473Dennis P. Hobbs (2004). Lean manufacturing implementation: a complete executionmanual for any size manufacturer. Boca Raton: Ross Publishing.James P. Womack及Daniel T. Jones (1996). Lean Think, Free Press.Khalil A. El-Namrouty, Mohammed S. AbuShaaban (2013). Seven wastes eliminationtargeted by lean manufacturing case study “gaza strip manufacturing firms’’. International Journal of Economics, Finance and Management Sciences, Vol. 1, No. 2: 68-80Mohd Javaid, Abid Haleem, Ravi Pratap Singh, Shanay Rab, Rajiv Suman, Shahbaz Khan (2021), Exploring Relationships Between Lean 4.0 and Manufacturing Industry, Industrial Robot, Vol. 49 Issue 3.Nitin S. Solke, Pritesh Shah, Ravi Sekhar & T. P. Singh (2022). Machine Learning-Based Predictive Modeling and Control of Lean Manufacturing in Automotive Parts Manufacturing Industry. Global Journal of Flexible Systems Management, volume 23, 89-112Oliver Nalbach, Christian Linn, Maximilian Derouet, and Dirk Werth(2018).Predictive Quality: Towards a New Understanding of Quality Assurance Using Machine Learning Tools. Business Information Systems, 30–42Sungyong Seo, Sercan Arik, Jinsung Yoon, Xiang Zhang, Kihyuk Sohn, Tomas Pfister(2021). Controlling Neural Networks with Rule Representations.In Advances in Neural Information Pro- cessing Systems.Zhiheng Huang, Wei Xu,Kai Yu (2015). Bidirectional LSTM-CRF Models for Sequence Tagging(Master thesis, Cornell University, New York, United States).Retrived from https://arxiv.org/abs/1508.01991二、 中文參考資料大野耐一（2016）。豐田生產方式。中國鐵道出版社王泰裕（2018）。專題報導—工業 4.0 使製造業升級。科學發展，第544期：04任苙萍（2019）。AI 是工業4.0 重頭戲 但「智慧製造」另有深意。智慧電子解決方案設計平台，產業特輯：21宋劭桓（2019）。基於深度學習之低功耗藍芽室內定位及其在機器人導航之應用。國立交通大學管理學院（電控工程研究所）碩士班碩士論文：03洪哲倫、張志宏、林宛儒（2019）。智慧機械專題—工業 4.0 與智慧製造的關鍵技術：工業物聯網與人工智慧。科儀新知，第221期：19-25洪哲倫（2020）。智慧製造的關鍵角色：工業大數據分析。工具機與控制器技術專輯，第444期：40-41許峻銘（2021）。基於深度學習之螺絲表面瑕疵檢測系統設計與實現。南臺科技大學（電子工程學系）碩士班碩士論文：1-75。陳韋儒（2018）。基於注意力機制長短期記憶深度學習之機器剩餘可用壽命預估。國立中央大學（資訊工程學系）碩士班碩士論文，頁1-42。劉仁傑（2015）。迎接工業4.0智慧型精實製造的挑戰。Machine Tool &Accessory，東海精實管理專欄：146-147劉瑞隆（2018）。專題報導—工業 4.0 使製造業升級。科學發展，第544期：17-20三、 網頁資料Christopher Olah (2015), Understanding LSTM Networks. Retrieved from http://colah. github.io/posts/2015-08-Understanding-LSTMs/.（搜尋日期：2022年5月30日）ED Sperling(2018), Deep Learning Spreads, Semiconductor Engineering Deepinsight for the tech industry, systems and design website:https://semiengineering.com/deep-learning-spreads/季平（2022）。提高產業韌性 智慧製造扮演關鍵角色。聯合新聞網網址https://udn.com/news/story/11726/6095658（搜尋日期：2022年5月30日）林均樺（2021）。學者觀點－從精實到數位化轉型之路。工商時報網址https://www.chinatimes.com/newspapers/20210906000117-260209?chdtv。（搜尋日期：2022年6月3日）黃博信（2020）。探討智慧製造發展趨勢。財團法人國家實驗研究院（科技政策研究與資訊中心），研究成果網站https://portal.stpi.narl.org.tw/index?p=article&id=4b1141ea74d7dcc40174f64de0372ac6（搜尋日期：2022年3月2日） | zh_TW |
| dc.identifier.doi (DOI) | 10.6814/NCCU202201083 | en_US |
