對精密加工而言，輪磨是最主要的加工方法，輪磨系統包含研磨機、操作條件、砂輪、工件、冷卻液等，為了達成工件的精度與加工效率要求，必須將輪磨系統中的相關變數組合最適化，其中選擇適合的砂輪規格對加工品質效率的影響甚大，例如在個案K公司的經驗中，用於線型滑軌鋼珠溝的成形研磨砂輪，經過砂輪規格最適化調整後，提升了30%的生產效率。
傳統砂輪規格推薦方式，銷售人員通常以自身經驗或詢問資深人員、工廠技術人員、公司內部技術資料等來得到推薦規格，但以上方法無統一標準以及涵蓋應用面不足，往往導致砂輪規格推薦不一致或不正確，並衍生後續交期過長、生產成本增加、以及推薦經驗知識無法累積等問題。另外由於輪磨系統的複雜性以及應用面甚廣，各砂輪製造商雖有自己的推薦規範，但通常只能提供粗略的建議，經常需要與客戶之間來回多次測試才能得到適用的規格，引發客戶抱怨風險。
為改善以上砂輪規格推薦系統的限制，本研究借助近幾年快速進步的人工智慧技術，以個案砂輪公司之專家知識、歷史案例之資料庫資料作為基礎，結合AI (Artificial Intelligence)系統以及專家系統(Expert System)等人工智慧工具，開發出整合專家知識、歷史案例、AI預測之混合式(Hybrid)砂輪規格推薦系統。本研究結果證明，本系統可以成功快速推薦最相似的砂輪規格，以及可隨資料庫增加不斷進步的AI推薦砂輪規格。

For precision machining, grinding is the primary processing method. A grinding system comprises a grinding machine, operating conditions, grinding wheel, workpiece, coolant, etc. To achieve the required precision and processing efficiency, it is necessary to optimize the relevant variables in the grinding system. The selection of an appropriate grinding wheel specification has a significant impact on processing quality and efficiency. For example, in the experience of Company K, after optimizing the grinding wheel specification for grinding ball grooves on linear slide rails, production efficiency increased by 30%.
Traditional recommendations for grinding wheel specifications are typically obtained through the experience of sales personnel, consulting senior personnel, factory technical staff, internal company technical data, etc. However, these methods lack uniform standards and insufficient coverage of applications, often resulting in inconsistent or incorrect grinding wheel recommendations. This leads to problems such as excessively long lead times, increased production costs, and the inability to accumulate knowledge from experienced personnel. Furthermore, due to the complexity and wide range of applications of grinding systems, although each grinding wheel manufacturer has its own recommended specifications, they can usually provide only rough suggestions. It often requires multiple rounds of testing with customers to obtain suitable specifications, cause customer complaint risk.
To improve the above grinding wheel specification recommendation systems, this study leverages the rapid advancements in artificial intelligence technology in recent years. Using the expert knowledge of Company K, and a database of historical cases as a foundation, this study integrates artificial intelligence tools, such as AI (Artificial Intelligence) and expert systems, to develop a hybrid grinding wheel specification recommendation system that combines expert knowledge (rule-based), historical cases (case-based), and AI predictions. The results of this study demonstrate that the system can successfully and rapidly recommend the most similar grinding wheel specifications, and it can continuously improve AI-recommended grinding wheel specifications as the database grows.

摘要 i
Abstract ii
目錄 iii
圖目錄 v
表目錄 vii
第一章 緒論 1
1.1研究背景與動機 1
1.2研究目標 4
1.3本論文結構 4
第二章 文獻回顧 6
2.1砂輪介紹 6
2.2輪磨系統介紹 7
2.3砂輪規格推薦文獻 11
2.4專家系統案例式推論(Case-based reasoning, CBR)文獻 14
2.5本研究與文獻之差異 17
第三章 研究方法 19
3.1資料準備 19
3.1.1數據收集 19
3.1.2數據清理 20
3.1.3數據轉換 23
3.1.4數據探索 26
3.2專家規則定義 28
3.2.1專家規則模組(Rule-based) 29
3.2.2相似案例模組(Case-based) 30
3.3 AI機器學習方法 32
3.4混合式模型 34
3.5系統開發方法 36
第四章 模型驗證 37
4.1查詢介面 37
4.2專家規則模組 38
4.3相似案例模組 41
4.4 AI推薦模組 43
4.4.1 AI模型參數實驗 43
4.4.2 AI模型表現分析 45
第五章 結論 49
5.1研究貢獻 49
5.2未來研究方向 50
參考文獻 51

齊人鵬、解文法(1991)。研磨技術。中國砂輪公司印行。
李家岩、洪佑鑫(2022)。製造數據科學。新北市：前程。
米田孝夫、山中将、嘉数侑昇(1992). ニューラルネットワークを用いた研削条件最適化に関する研究, 精密工学会誌, 58(10), 85-90.
Aamodt, A. , Plaza,E. (1994). Case-based reasoning: foundational issues, methodological variations and system approaches. AI Communications, 7(1), 39-59.
Bhowmika, S. , Naik, R. (2018). Selection of abrasive materials for manufacturing grinding wheels. International Conference on Advanced Materials and Applications (ICAMA 2016). 2860–2864.
Fernández,R., Iriarte, A., Puerto, P., Gallego, I., & Arrazola, P. J. (2011). Analysis of the behavior of grinding wheels in surface grinding. The 4th Manufacturing Engineering Society International Conference (Mesic 2011), Cadiz, Spain.
JIS (2014). Recommendation of grinding wheels. JIS B4051: 2014
Kodama,H., Uotani I.& Ohashi,K. (2019). Decision support system for principal factors of grinding wheel using data mining methodology. International Journal of Abrasive Technology, 9(2), 89-98.
Kolodner, J. L. (1983). Reconstructive memory: A computer model. Cognitive Science, 7(4), 281-328.
Lahitani, A. R., Permanasari, A. E., & Setiawan, N. A. (2016). Cosine similarity to determine similarity measure: study case in online essay assessment. 2016 4th International Conference on Cyber and IT Service Management, Bandung, Indonesia.
Liao, T. W. , Zhang, Z. (1998). Similarity measures for retrieval in case-based reasoning systems. Applied Artificial Intelligence, 12, 267-288.
Li, B. , Han, L. (2013). Distance weighted cosine similarity measure for text classification. 14th International Conference, IDEAL 2013, Hefei, China.
Li, Y., Mills, B., & , Rowe, W. B. (1997). An intelligent system for selection of grinding wheels. Proceedings of the Institution of Mechanical Engineers, Part B, 635-641.
Li, Y., Rowe, W. B., & Mills, B. (1999). Study and selection of grinding conditions Part 1: grinding conditions and selection strategy. Proceedings of the Institution of Mechanical Engineers, Part B, 119-129.
Li, Y., Rowe, W. B., Chen, X. , & Mills, B. (1999). Study and selection of grinding conditions Part 2: a hybrid intelligent system for selection of grinding conditions. Proceedings of the Institution of Mechanical Engineers, Part B, 131-142.
Maity, S. R. , Chakraborty, S. (2013). Grinding wheel abrasive material selection using fuzzy TOPSIS method. Materials and Manufacturing Processes, 28, 408–417.
Midha, P. S. , Zhu, C. B., & Trmal, G. J. (1990). An expert system for grinding wheel selection for cylindrical grinding operations. Proceedings of 6th Int. Conf. on Computer Aided Production Engineering, London.
Ontañón, S. (2020). An overview of distance and similarity functions for structured data. Artificial Intelligence Review. 53, 5309–5351.
Rowe, W. B., Li, Y. (1994). Applications of artificial intelligence in grinding. Annals of the ClRP, 43(2), 521-531.
Schank, R.C. & Abelson, R.P. (1977). Scripts, plans, goals and understanding. Erlbaum, Hillsdale, New Jersey, US.
Schank, R. (1982). Dynamic memory: a theory of reminding and learning in computers and people. Cambridge University Press, Cambridge, UK.
Shore, P., Billing, O., & Puhasmagi, V. (2004). A standard grinding wheel assessment method to support a sophisticated grinding knowledge based system. Key Engineering Materials, 257, 285-290.
Singhal, A. (2001). Modern information retrieval: a brief overview. Bulletin of the Technical Committee on Data Engineering, 24(4), 35-43.
Trmal, G. J. , Zhu, C. B., & Midha, P. S. (1992). An expert system for grinding process Optimization. Journal of Materials Processing Technology. 33, 507-517.
Watson, I. , Marir, F. (1994). Case-based reasoning: a review. The Knowledge Engineering Review, 9(4), 1-34.