臺灣博碩士論文加值系統

在金屬加工產業中，高週波熱處理是關鍵技術之一。本研究透過智慧製造轉型，建立機台參數推薦系統及結果模擬工具，旨在降低設定機台參數的專業門檻，允許操作者模擬不同參數設定的影響，進而減少實驗成本並提升生產效率。應用機器學習技術，研究實現高週波熱處理產品加工後硬度預測模型，幫助操作者判斷產品是否符合質量標準。面對傳統設備規劃的局限，本研究比較了多種線性回歸、機器學習模型，並採用超參數優化演算法進一步優化模型性能，使用留一法交叉驗證(Leave-One-Out Cross-Validation, LOOCV)來確保模型的泛化能力。實驗結果顯示，在 T07 長軸心的預測研究中，使用 AdaBoost(Adaptive Boosting)結合隨機搜索達到最佳效果，MAE(Mean Absolute Error)為1.292。在 T08 排齒的預測中，決策樹模型的 MAE 最小為0.8333。而在 T12 回火設備的預測中，使用AdaBoost 和 Optuna 的模型達到 MAE 最小值為0.6605。這些成果為金屬熱處理產業提供了有效的智慧化生產工具，有望顯著提升效率和質量。

In the metal processing industry, high-frequency heat treatment is a key technology in the metal processing industry. This study aims to enhance smart manufacturing by creating a machine parameter recommendation system and simulation tools. These tools lower the professional threshold for parameter settings, allowing operators to simulate outcomes, thus reducing costs and improving efficiency. Machine learning techniques were used to develop a hardness prediction model for products post-treatment. The study compared various linear regression and machine learning models, employing hyperparameter optimization and Leave-One-Out Cross-Validation for performance enhancement. Results showed that the AdaBoost model with random search achieved the best MAE( Mean Absolute Error) of 1.292 for T07 shafts, decision tree model had an MAE of 0.8333 for T08 gears, and AdaBoost with Optuna reached an MAE of 0.6605 for T12 tempering equipment. These findings provide valuable tools for the metal heat treatment industry, potentially improving both efficiency and product quality.

目錄
摘要 ......................................................................................................... I
致謝 ....................................................................................................... III
目錄 ....................................................................................................... IV
圖目錄 ................................................................................................... VI
表目錄 ................................................................................................. VIII
名詞縮寫表 ............................................................................................ X
第一章 緒論 ........................................................................................ 1
1.1 研究背景與動機 .............................................................................................. 1
1.1.1 熱處理之方法種類 .................................................................................. 1
1.1.2 高週波熱處理之方法 .............................................................................. 4
1.1.3 產業需求及挑戰 ...................................................................................... 4
1.2 研究流程 .......................................................................................................... 6
1.3 論文架構 .......................................................................................................... 9
第二章 文獻探討 .............................................................................. 10
2.1 金屬加工熱處理現況分析 ............................................................................ 11
2.2 金屬加工業導入智慧製造案例 .................................................................... 12
2.3 資料前處理(Data Preprocessing) ................................................................... 12
2.3.1 為什麼需要資料前處理？ .................................................................... 12
2.3.2 資料前處理之步驟 ................................................................................ 13
2.4 線性回歸模型 ................................................................................................ 15
2.4.1 線性回歸(Liner Regression) ................................................................... 15
2.4.2 嶺回歸(Ridge Regression) ...................................................................... 17
2.4.3 套索回歸(Lasso Regression) .................................................................. 18
2.4.4 支持向量回歸(Support Vector Regression, SVR) ................................... 19
2.5 機器學習模型 ................................................................................................ 20
2.5.1 決策樹(Decision Tree) ............................................................................ 20
2.5.2 隨機森林(Random Forest) ..................................................................... 22
2.5.3 自適應增強(Adaptive Boosting ,AdaBoost) ........................................... 24
2.5.4 極限梯度提升(Extreme Gradient Boosting ,XGBoost) ........................... 26
2.6 超參數優化方法 ............................................................................................ 29
2.6.1 Grid Search ............................................................................................. 29
2.6.2 Random Search ...................................................................................... 30
2.6.3 Bayesian Optimization ............................................................................ 32
2.6.4 Optuna .................................................................................................... 33
2.7 留一法交叉驗證(Leave-One-Out Cross-Validation, LOOCV) ......................... 35
2.8 模型評估驗證指標 ........................................................................................ 36
第三章 研究架構與方法 .................................................................. 40
3.1 數據蒐集 ........................................................................................................ 41
3.2 資料前處理 .................................................................................................... 43
3.3 模型訓練與優化 ............................................................................................ 51
3.4 性能評估 ........................................................................................................ 52
第四章 研究結果與比較 .................................................................. 53
4.1 自適應超參數優化前模型比較 .................................................................... 54
4.1.1 機器學習方法 ........................................................................................ 54
4.1.2 線性回歸方法 ........................................................................................ 56
4.2 自適應超參數優化後模型比較 .................................................................... 58
4.2.1 XGBoost .................................................................................................. 58
4.2.2 AdaBoost ................................................................................................ 60
4.3 總比較 ............................................................................................................ 62
4.4 網頁呈現 ........................................................................................................ 65
第五章 結論 ...................................................................................... 68
5.1 結論 ................................................................................................................ 68
5.2 未來展望 ........................................................................................................ 69
參考文獻 .............................................................................................. 70

[1] Banerjee, M. K. (2017). 2.1 Fundamentals of heat treating metals and alloys.
Comprehensive materials finishing, 1-49.
[2] Munusamy, S., & Jerald, J. (2023). Effect of in-Situ Intrinsic Heat Treatment in
Metal Additive Manufacturing: A Comprehensive Review. Metals and Materials
International, 29(12), 3423-3441
[3] Gajdzik, B., Wolniak, R., & Grebski, W. W. (2023). Electricity and heat demand in
steel industry technological processes in Industry 4.0 conditions. Energies, 16(2),
787.
[4] 陳彥儒, 許曉弘, 陳其褘, & 張智銘. (2020). 金屬材料熱處理模擬分析技術發
展現況. 機械工業雜誌, (453), 42-47.
[5] 蕭博仁(2021)。應用機器學習於金屬銲接參數最佳化。﹝碩士論文。華梵大
學﹞臺灣博碩士論文知識加值系統。 https://hdl.handle.net/11296/v4cs86。
[6] 吳姎恂(2020)。應用不同深度學習工具以提高金屬加工產品瑕疵檢測之影像
辨識成功率。﹝碩士論文。明志科技大學﹞臺灣博碩士論文知識加值系統。
https://hdl.handle.net/11296/axv5ga。
[7] Han, J., Pei, J., & Tong, H. (2022). Data mining: concepts and techniques. Morgan
kaufmann.
[8] Chen, M. S., Han, J., & Yu, P. S. (1996). Data mining: an overview from a database
perspective. IEEE Transactions on Knowledge and data Engineering, 8(6), 866-
883.
[9] Weisberg, S. (2005). Applied linear regression (Vol. 528). John Wiley & Sons.
[10] Hoerl, A. E., & Kennard, R. W. (1970). Ridge regression: Biased estimation for
nonorthogonal problems. Technometrics, 12(1), 55-67.
[11] Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of
the Royal Statistical Society Series B: Statistical Methodology, 58(1), 267-288.
[12] Drucker, H., Burges, C. J., Kaufman, L., Smola, A., & Vapnik, V. (1996). Support
71
vector regression machines. Advances in neural information processing systems, 9.
[13] Timofeev, R. (2004). Classification and regression trees (CART) theory and
applications. Humboldt University, Berlin, 54.
[14] Cutler, A., Cutler, D. R., & Stevens, J. R. (2012). Random forests. Ensemble
machine learning: Methods and applications, 157-175.
[15] Freund, Y., & Schapire, R. E. (1996, July). Experiments with a new boosting
algorithm. In icml (Vol. 96, pp. 148-156).
[16] Chen, T., & Guestrin, C. (2016, August). Xgboost: A scalable tree boosting system.
In Proceedings of the 22nd acm sigkdd international conference on knowledge
discovery and data mining (pp. 785-794).
[17] Bergstra, J., & Bengio, Y. (2012). Random search for hyper-parameter
optimization. Journal of machine learning research, 13(2).(random search)
[18] Zou, M., Jiang, W. G., Qin, Q. H., Liu, Y. C., & Li, M. L. (2022). Optimized
XGBoost model with small dataset for predicting relative density of Ti-6Al-4V parts
manufactured by selective laser melting. Materials, 15(15), 5298.
[19] Famili, A., Shen, W. M., Weber, R., & Simoudis, E. (1997). Data preprocessing and
intelligent data analysis. Intelligent data analysis, 1(1), 3-23.
[20] Freund, Y., & Schapire, R. E. (1997). A decision-theoretic generalization of on-line
learning and an application to boosting. Journal of computer and system
sciences, 55(1), 119-139.
[21] 陳婷文. (2020). Lasso 迴歸於 可詮釋預測分析: 強階層與樹狀結構. 政治大學
資訊管理學系學位論文, 2020, 1-43.
[22] Tao, F., Qi, Q., Liu, A., & Kusiak, A. (2018). Data-driven smart
manufacturing. Journal of Manufacturing Systems, 48, 157-169.