臺灣博碩士論文加值系統

SKD11合金工具鋼是一種高碳高鉻合金鋼，這種鋼材經過熱處理後，硬度可達到HRC58-62度，亦即它具有高強度、高硬度、高韌度、表面抗磨耗能力、較高的疲勞強度和抗變形等優越特性，在工業中，它主要廣泛應用於製造沖壓模、塑膠模等多種模具。 為了改善SKD11的切削性能、加工效率以及刀具使用壽命，嘗試尋求改進製程、探索輔助切削方法和優化參數組合等方法，以使這種廣泛應用的材料之加工更具經濟效益。
本文研究不同輔助方式對SKD11銑削的影響，並找出最佳的輔助方式和切削參數。輔助方式包括旋轉超音波輔助、雙軸超音波輔助以及旋轉超音波結合雙軸超音波混合輔助。在實驗中，以碳化鎢鍍上TiAlN的銑刀進行銑削。研究分為二個階段。第一階段研究比較兩種不同參數和三種輔助方式之間的關係。第二階段基於第一階段實驗的結果，選擇最佳輔助方式，並通過調整主軸轉速、進給速率和徑向切深等加工參數，以田口法直交表進行實驗設計的同時，引入雷射加熱輔助技術，對SKD11進行側邊銑削實驗。在兩個階段的實驗中，使用動力計來監測切削力的變化，使用工具顯微鏡觀察SKD11的邊緣形貌和銑刀的磨耗情況，並使用表面粗糙度儀來測量表面粗糙度。最終，我們以達到所需表面粗糙度為目標，同時考慮到刀具壽命的限制條件，運用田口品質法的望小特性，得出最佳的加工參數組合。第二階段的實驗結果顯示，與無輔助相比，使用超音波輔助可以獲得較好的表面粗糙度和邊緣形貌，其中旋轉超音波的效果優於雙軸與雙軸結合旋轉超音波輔助。在延續最佳輔助方式的基礎上，引入雷射輔助後，旋轉超音波結合雷射輔助在表面粗糙度與切削力方面表現更好，切削力降低了16％，表面粗糙度提高了30％。因此，結合旋轉超音波和雷射輔助被證實是本研究的最佳輔助方式。
根據第二階段的實驗結果，以旋轉超音波結合雷射輔助進行重複銑削實驗，蒐集切削力數據與刀具磨耗數據，建立基於集成學習的刀具磨耗預測模型，這種方法利用多種機器學習回歸模型，包括決策樹 (Decision Tree)、隨機森林 (Random Forest)、極端隨機樹 (Extremely Randomized Tree)、LGBM (Light Gradient Boosting Machine)、XGBoost (Extreme Gradient Boosting) 、AdaBoost (Adaptive Boosting)、隨機梯度下降(Stochastic Gradient Descent)、支持向量回歸(Support Vector Regression)、線性支持向量回歸(Linear Support Vector Regression)和多層感知器(Multilayer Perceptron)，在經過超參數調整後，基於上述十個模型進行集成投票回歸預測。實驗結果顯示集成投票回歸模型超越單一機器學習回歸模型之成效，該模型在決定係數(R2)達 0.94576、均方根誤差(RMSE)達0.24348、均方誤差(MSE)達0.05928與平方絕對誤差(MAE)達0.18182的成效，因此集成學習之方法用來監測刀具磨耗為可行且有效的方法。

The SKD11 alloy tool steel is a high-carbon, high-chromium alloy steel. After undergoing heat treatment, this steel can achieve a hardness of HRC58-62. This means it possesses high strength, high hardness, high toughness, surface wear resistance, higher fatigue strength, and resistance to deformation. In the industrial sector, it's extensively used for manufacturing stamping dies, plastic molds, and various other molds. To enhance the cutting performance, machining efficiency, and cutting-tool life of SKD11, efforts have been channeled into refining the manufacturing process, exploring assisted cutting methods, and optimizing parameter combinations, aiming to make the machining of this widely-used material more economically efficient.
This study investigates the influence of different assisted methods on the milling of SKD11 and determines the optimal assisted method and cutting process parameters. The assisted methods examined include rotary ultrasonic assistance, dual-axis ultrasonic assistance, and a combination of rotary and dual-axis ultrasonic assistance. In the experiments, carbide cutters coated with TiAlN were used for side milling. The study was divided into two phases. The first phase compared the relationships between two different process parameters and the three assisted methods. In the second phase, based on the results of the first phase, the best assisted method from the first phase was chosen. By adjusting the spindle speed, feed rate, and radial cutting depth, experiments were designed using the Taguchi orthogonal array method. Concurrently, laser heating assisted technology was introduced for side milling of SKD11. Throughout both experimental phases, a dynamometer was employed to monitor changes in cutting force, a tool microscope was used to observe the edge morphology of SKD11 and the wear condition of the milling cutter, and a surface roughness instrument was used to measure the surface roughness. Ultimately, the goal was to achieve the desired surface roughness, while considering the constraints of cutting-tool life. The results from the second phase showed that compared to no assistance, using ultrasonic assistance achieved a better surface roughness and edge morphology. Among these methods, rotary ultrasonic assistance prevails over dual-axis ultrasonic assistance and the combined rotary-dual-axis ultrasonic assisted method. Building on the optimal assisted method, introducing laser assistance combined with rotary ultrasonic assistance resulted in even better surface roughness and cutting force outcomes, with a 16% reduction in cutting force and a 30% improvement in surface roughness. As a result, the combination of rotary ultrasonic and laser assistance was proven to be the most effective assisted method in this study.
Based on the outcomes of the second phase, repeated milling experiments using rotary ultrasonic combined with laser assistance were conducted. Data on cutting force and cutting-tool wear were collected, leading to the establishment of a cutting-tool wear prediction model based on ensemble learning. This approach utilized multiple machine learning regression models, including Decision Tree, Random Forest, Extremely Randomized Tree, Light Gradient Boosting Machine, Extreme Gradient Boosting, Adaptive Boosting, Stochastic Gradient Descent, Support Vector Regression, Linear Support Vector Regression, and Multilayer Perceptron. After hyperparameter tuning, ensemble voting regression predictions were made based on these ten models. The results indicated that the ensemble voting regression model surpassed individual machine learning regression models in performance, achieving a coefficient of determination (R2) of 0.94576, a root mean square error (RMSE) of 0.24348, a mean squared error (MSE) of 0.05928, and a mean absolute error (MAE) of 0.18182. Therefore, using ensemble learning as a method to monitor tool wear is both feasible and effective.

摘要 ...i
Abstract ...iii
誌謝 ...v
目錄 ...vi
表目錄 ...viii
圖目錄 ...ix
第一章 前言 ...1
1.1 研究背景 ...1
1.2 研究動機與目的 ...2
1.3 文獻回顧 ...3
1.4 論文架構 ...6
第二章 理論基礎 ...8
2.1 超音波輔助技術 ...8
2.2 雷射輔助切削技術 ...8
2.3 田口實驗法 ...9
2.4 刀具磨耗 ...11
2.5 機器學習 ...12
2.6 集成學習 ...12
2.7 回歸模型 ...13
第三章 實驗規劃與方法 ...15
3.1 工件與刀具 ...16
3.2 實驗儀器設備與規格 ...17
3.3 銑削實驗架構 ...27
3.3.1 第一階段銑削實驗 ...28
3.3.2 第二階段銑削實驗 ...29
3.3.3 資料收集實驗 ...30
3.4 機器學習架構 ...31
第四章 資料處理 ...32
4.1 資料清理 ...32
4.2 特徵提取 ...33
4.3 資料標準化與特徵選擇 ...35
第五章 結果與討論 ...36
5.1 表面粗糙度比較 ...36
5.1.1 第一階段實驗 ...36
5.1.2 第二階段實驗 ...37
5.2 切削力比較 ...39
5.2.1 第一階段實驗 ...39
5.2.2 第二階段實驗 ...39
5.3 SKD11邊緣形貌比較 ...44
5.4 刀具磨耗比較 ...48
5.5 機器學習結果比較 ...52
5.5.1 機器學習結果比較 ...53
5.5.2 機器學習與集成學習結果比較 ...56
第六章 結論與建議 ...58
參考文獻 ...60
Extended Abstract ...63

[1] J. V. Abellan-Nebot and F. Romero Subirón, 2010, “A review of machining monitoring systems based on artificial intelligence process models,” The International Journal of Advanced Manufacturing Technology, Vol. 47, No. 1, pp. 237-257.
[2] S. Y. Liang, R. L. Hecker and R. G. Landers, 2004, “Machining Process Monitoring and Control: The State-of-the-Art,” Journal of Manufacturing Science and Engineering, Vol. 126, No. 2, pp. 297-310.
[3] M. Anderson, R. Patwa and Y. C. Shin, 2006, “Laser-assisted machining of Inconel 718 with an economic analysis,” International Journal of Machine Tools and Manufacture, Vol. 46, No. 14, pp. 1879-1891.
[4] M. Mejbel, M. Khalaf and A. Kwad, 2021, “Improving the Machined Surface of AISI H11 Tool Steel in Milling Process,” Journal of Mechanical Engineering Research and Developments, Vol. 44, pp. 58-68.
[5] M. Y. Tsai, C. T. Chang and J. K. Ho, 2016, “The Machining of Hard Mold Steel by Ultrasonic Assisted End Milling,” Applied Sciences, Vol. 6, No. 11, p. 373.
[6] H. Ding, R. Ibrahim, K. Cheng and S.-J. Chen, 2010, “Experimental study on machinability improvement of hardened tool steel using two dimensional vibration-assisted micro-end-milling,” International Journal of Machine Tools and Manufacture, Vol. 50, No. 12, pp. 1115-1118.
[7] G. S. Ghule, S. Sanap, S. Adsul, S. Chinchanikar and M. Gadge, 2022, “Experimental investigations on the ultrasonic vibration-assisted hard turning of AISI 52100 steel using coated carbide tool,” Materials Today: Proceedings, Vol. 68, pp. 2093-2098.
[8] X. Zhang, A. Senthil Kumar, M. Rahman, C. Nath and K. Liu, 2011, “Experimental study on ultrasonic elliptical vibration cutting of hardened steel using PCD tools,” Journal of Materials Processing Technology, Vol. 211, No. 11, pp. 1701-1709.
[9] X. Chen, J. Xu and Q. Xiao, 2015, “Cutting performance and wear characteristics of Ti(C,N)-based cermet tool in machining hardened steel,” International Journal of Refractory Metals and Hard Materials, Vol. 52, pp. 143-150.
[10] H. Saito, H. Jung and E. Shamoto, 2016, “Elliptical vibration cutting of hardened die steel with coated carbide tools,” Precision Engineering, Vol. 45, pp. 44-54.
[11] M. Kumar and S. N. Melkote, 2012, “Process capability study of laser assisted micro milling of a hard-to-machine material,” Journal of Manufacturing Processes, Vol. 14, No. 1, pp. 41-51.
[12] C. Brecher, M. Emonts, C.-J. Rosen and J.-P. Hermani, 2011, “Laser-assisted Milling of Advanced Materials,” Physics Procedia, Vol. 12, pp. 599-606.
[13] W.-S. Woo and C.-M. Lee, 2015, “A study of the machining characteristics of AISI 1045 steel and Inconel 718 with a cylindrical shape in laser-assisted milling,” Applied Thermal Engineering, Vol. 91, pp. 33-42.
[14] S. Xavierarockiaraj and P. Kuppan, 2014, “Investigation of Cutting Forces, Surface Roughness and Tool Wear during Laser Assisted Machining of SKD11Tool Steel,” Procedia Engineering, Vol. 97, pp. 1657-1666.
[15] J. Vivancos, C. J. Luis, J. A. Ortiz and H. A. González, 2005, “Analysis of factors affecting the high-speed side milling of hardened die steels,” Journal of Materials Processing Technology, Vol. 162-163, pp. 696-701.
[16] C. Y. Wang, Y. X. Xie, Z. Qin, H. S. Lin, Y. H. Yuan and Q. M. Wang, 2015, “Wear and breakage of TiAlN- and TiSiN-coated carbide tools during high-speed milling of hardened steel,” Wear, Vol. 336-337, pp. 29-42.
[17] 簡鴻喨，2007，硬化熱作模具鋼之高速銑削特性研究，國立臺灣海洋大學機械與機電工程學系碩士論文
[18] R. C. Dewes, E. Ng, K. S. Chua, P. G. Newton and D. K. Aspinwall, 1999, “Temperature measurement when high speed machining hardened mould/die steel,” Journal of Materials Processing Technology, Vol. 92-93, pp. 293-301.
[19] Z. Pu and A. Singh, 2013, “High speed ball nose end milling of hardened AISI A2 tool steel with PCBN and coated carbide tools,” Journal of Manufacturing Processes, Vol. 15, No. 4, pp. 467-473.
[20] F. Gong, J. Zhao, Y. Jiang, H. Tao, Z. Li and J. Zang, 2017, “Fatigue failure of coated carbide tool and its influence on cutting performance in face milling SKD11 hardened steel,” International Journal of Refractory Metals and Hard Materials, Vol. 64, pp. 27-34.
[21] H. Sun, J. Zhang, R. Mo and X. Zhang, 2020, “In-process tool condition forecasting based on a deep learning method,” Robotics and Computer-Integrated Manufacturing, Vol. 64, p. 101924.
[22] K. Zhu, G. Li and Y. Zhang, 2019, “Big Data Oriented Smart Tool Condition Monitoring System,” IEEE Transactions on Industrial Informatics, Vol. PP, pp. 1-1.
[23] H. Guo and K. P. Zhu, 2021, “Attention-based dual-scale hierarchical LSTM for tool wear monitoring,” Manufacturing Letters, Vol. 29, pp. 99-103.
[24] X. Xu, Z. Tao, W. Ming, Q. An and M. Chen, 2020, “Intelligent monitoring and diagnostics using a novel integrated model based on deep learning and multi-sensor feature fusion,” Measurement, Vol. 165, p. 108086.
[25] Z. M. Çınar, A. Abdussalam Nuhu, Q. Zeeshan, O. Korhan, M. Asmael and B. Safaei, 2020, “Machine Learning in Predictive Maintenance towards Sustainable Smart Manufacturing in Industry 4.0,” Sustainability, Vol. 12, No. 19, p. 8211.
[26] I. H. Sarker, 2021, “Machine Learning: Algorithms, Real-World Applications and Research Directions,” SN Computer Science, Vol. 2, No. 3, p. 160.
[27] O. Sagi and L. Rokach, 2018, “Ensemble learning: A survey,” WIREs Data Mining and Knowledge Discovery, Vol. 8, No. 4, p. e1249.
[28] J. R. Quinlan, 1986, “Induction of decision trees,” Machine Learning, Vol. 1, No. 1, pp. 81-106.
[29] L. Breiman, 2001, “Random Forests,” Machine Learning, Vol. 45, No. 1, pp. 5-32.
[30] P. Geurts, D. Ernst and L. Wehenkel, 2006, “Extremely randomized trees,” Machine Learning, Vol. 63, No. 1, pp. 3-42.
[31] T. Chen and C. Guestrin, 2016, XGBoost: A Scalable Tree Boosting System, presented at the Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, California, USA,
[32] G. Ke, Q. Meng, T. Finley, T. Wang, W. Chen, W. Ma, Q. Ye and T.-Y. Liu, 2017, LightGBM: A Highly Efficient Gradient Boosting Decision Tree,
[33] L. Bottou, "Large-Scale Machine Learning with Stochastic Gradient Descent," in Proceedings of COMPSTAT'2010, Heidelberg, Y. Lechevallier and G. Saporta, Eds., 2010// 2010: Physica-Verlag HD, pp. 177-186.
[34] Y. Freund and R. E. Schapire, "A Short Introduction to Boosting," 1999.
[35] C. Cortes and V. Vapnik, 1995, “Support-vector networks,” Machine Learning, Vol. 20, No. 3, pp. 273-297.
[36] M. A. Elbestawi, T. A. Papazafiriou and R. X. Du, 1991, “In-process monitoring of tool wear in milling using cutting force signature,” International Journal of Machine Tools and Manufacture, Vol. 31, No. 1, pp. 55-73.
[37] E. Kuljanic and M. Sortino, 2005, “TWEM, a method based on cutting forces—monitoring tool wear in face milling,” International Journal of Machine Tools and Manufacture, Vol. 45, No. 1, pp. 29-34.
[38] P. Gierlak, A. Burghardt, D. Szybicki, M. Szuster and M. Muszyńska, 2017, “On-line manipulator tool condition monitoring based on vibration analysis,” Mechanical Systems and Signal Processing, Vol. 89, pp. 14-26.
[39] Y. Qin, X. Liu, C. Yue, M. Zhao, X. Wei and L. Wang, 2023, “Tool wear identification and prediction method based on stack sparse self-coding network,” Journal of Manufacturing Systems, Vol. 68, pp. 72-84.
[40] X. Mao, F. Zhang, G. Wang, Y. Chu and K. Yuan, 2021, “Semi-random subspace with Bi-GRU: Fusing statistical and deep representation features for bearing fault diagnosis,” Measurement, Vol. 173, p. 108603.