現在的金屬積層製造系統都以高能量源來熔融金屬材料，最常見的高能量源是雷射與電子束，而這些高能量源被認為是穩定的能源，但是最大的缺點是其高耗能系統且成本昂貴。而本研究所使用的金屬電弧積層製造(Wire and arc additive manufacturing, WAAM)的特點為成本低廉，但不穩定的能量控制系統使得此製造技術具有精度和表面不平整的問題。
本研究架設監控視覺系統以數值控制平台(Computer Numerical Control, CNC)在1.5 mm/s速度移動下連續拍攝熔融沉積成型結果，藉由模板匹配影像處理技術合成約1300張連續影像，並利用邊緣檢測演算法提取不同進料參數下的金屬高度輪廓，而此輪廓資訊將會用於強化學習模型之訓練，以取得優秀的進料參數策略來改進表面高低起伏的情況。
本研究整合電腦視覺、影像處理及強化學習技術，來學習最佳進料參數策略，並於製程過程中使用該組參數策略，以降低製程結果的外型變化量。本研究使用Q-learning演算法的參數策略決策系統，運用學習的方式在少數訓練回合內，有效率的找出能將高低起伏控制在0.29 mm的進料參數策略，此方法不需耗費大量實驗時間去尋找好的製程參數。

Currently, metal additive manufacturing systems can melt metal materials using high-energy sources. The most common high-energy sources such as lasers and electron beams are considered the stable energy sources, but the biggest weakness of these sources is the costly and high energy consumption system. In this research, the wire and arc additive manufacturing (WAAM) is used which characterized by low cost. However, the imperfect energy control system makes this manufacturing technology have the following shortcomings: accuracy and surface irregularities.
In this research, the monitoring vision system is set up to capture fused deposition images while the CNC platform is moving at speed 1.5 mm/s. About 1300 continuous images are synthesized by template matching, and then the full images under different feeding rates will be extracted contour information by using edge detection. The contour information will be used for training reinforcement learning model to obtain a good feeding rate strategy to improve the profile accuracy.
Therefore, this research integrates computer vision, image processing, and reinforcement learning technologies to learn the best feeding rate parameter strategy. Then use this set of parameter strategy in the process to reduce the appearance change of the printed result. This research uses a parameter strategy decision-making method based on the Q-learning algorithm to efficiently find out the good feeding rate strategy that can reduce appearance change to 0.29 mm in few training episodes. This method does not need to spend a lot of experimental time to find good process parameters.

摘要 I
Abstract II
目錄 VI
圖目錄 X
表目錄 XV
第一章 緒論 1
1.1 前言 1
1.2 研究動機 3
第二章 文獻回顧 6
2.1 金屬積層製造 6
2.2 金屬積層製造監控技術 13
2.2.1 接觸式監控系統 14
2.2.2 非接觸式監控系統 15
2.3 影像處理 20
2.3.1 模板匹配 20
2.3.2 影像降噪 22
2.3.3 形態學處理 25
2.3.4 影像分割 27
2.3.5 邊緣檢測 30
2.4 機器學習 32
2.4.1 機器學習簡介 32
2.4.2 強化學習 33
2.4.3 機器學習應用於積層製造 37
第三章 研究方法 39
3.1 金屬材料成型實驗架構 42
3.1.1 金屬材料 42
3.1.2 印製基板夾具設計 43
3.1.3 電漿電弧系統架構 45
3.2 監控環境建構 47
3.2.1 視覺監控 47
3.2.2 影像擷取系統建置 49
3.3 影像處理 51
3.3.1 影像合成 52
3.3.2 影像降噪 53
3.3.3 邊緣檢測 54
3.4 強化學習 55
第四章 研究結果與討論 57
4.1 進料機構造成之溢料 57
4.2 不鏽鋼熔融成型結果 59
4.3 影像處理結果分析 60
4.3.1 影像合成方法探討 60
4.3.2 邊緣檢測探討 61
4.3.3 高度輪廓資訊的顯示與紀錄 63
4.4 中間段落強化學習結果討論 64
4.4.1 印製學習中間段落的數據 64
4.4.2 中間段落之強化學習結果 67
4.5 頭尾端製程結果討論 73
4.5.1 進料前預熱造成的影響 73
4.5.2 後尾抽料速度造成的影響 74
4.6 前頭端強化學習結果討論 76
4.6.1 印製學習前頭端的數據 76
4.6.2 前頭端的獎懲函數設計 77
4.6.3 前頭段之強化學習結果 79
4.7 第二層疊層結果 80
第五章 結論與未來展望 82
5.1 結論 82
5.2 未來展望 83

[1] Q. Chen, G. Guillemot, C.-A. Gandin, and M. Bellet, "Three-dimensional finite element thermomechanical modeling of additive manufacturing by selective laser melting for ceramic materials," Additive Manufacturing, vol. 16, pp. 124-137, 2017.
[2] B. Derby, "Additive manufacture of ceramics components by inkjet printing," Engineering, vol. 1, no. 1, pp. 113-123, 2015.
[3] J.-P. Kruth, M.-C. Leu, and T. Nakagawa, "Progress in additive manufacturing and rapid prototyping," Cirp Annals, vol. 47, no. 2, pp. 525-540, 1998.
[4] B. Lu, D. Li, and X. Tian, "Development trends in additive manufacturing and 3D printing," Engineering, vol. 1, no. 1, pp. 85-89, 2015.
[5] J. Gu, B. Cong, J. Ding, S. W. Williams, and Y. Zhai, "Wire+ arc additive manufacturing of aluminium," in Proceedings of the 25th Annual International Solid Freeform Fabrication Symposium, Austin, TX, USA, pp. 4-6, 2014.
[6] D. Ding, Z. Pan, D. Cuiuri, and H. Li, "Wire-feed additive manufacturing of metal components: technologies, developments and future interests," The International Journal of Advanced Manufacturing Technology, vol. 81, no. 1-4, pp. 465-481, 2015.
[7] F. Martina, P. A. Colegrove, S. W. Williams, and J. Meyer, "Microstructure of interpass rolled wire+ arc additive manufacturing Ti-6Al-4V components," Metallurgical and Materials Transactions A, vol. 46, no. 12, pp. 6103-6118, 2015.
[8] J. Hönnige, P. A. Colegrove, S. Ganguly, E. Eimer, S. Kabra, and S. Williams, "Control of residual stress and distortion in aluminium wire+ arc additive manufacture with rolling," Additive Manufacturing, vol. 22, pp. 775-783, 2018.
[9] J.R. Hönnige, P.A. Colegrove, B. Ahmad, M.E. Fitzpatrick, S. Ganguly, T.L. Lee, S.W. Williams, "Residual stress and texture control in Ti-6Al-4V wire+ arc additively manufactured intersections by stress relief and rolling," Materials & Design, vol. 150, pp. 193-205, 2018.
[10] F. Montevecchi, G. Venturini, N. Grossi, A. Scippa, and G. Campatelli, "Idle time selection for wire-arc additive manufacturing: A finite element-based technique," Additive Manufacturing, vol. 21, pp. 479-486, 2018.
[11] F. Li, S. Chen, J. Shi, Y. Zhao, and H. Tian, "Thermoelectric cooling-aided bead geometry regulation in wire and arc-based additive manufacturing of thin-walled structures," Applied Sciences, vol. 8, no. 2, p. 207, 2018.
[12] K. Zeng, D. Pal, and B. Stucker, "A review of thermal analysis methods in laser sintering and selective laser melting," in Proceedings of Solid Freeform Fabrication Symposium Austin, TX, vol. 60, pp. 796-814, 2012.
[13] G. Bi, A. Gasser, K. Wissenbach, A. Drenker, and R. Poprawe, "Identification and qualification of temperature signal for monitoring and control in laser cladding," Optics and Lasers in Engineering, vol. 44, no. 12, pp. 1348-1359, 2006.
[14] S. Kleszczynski, J. Zur Jacobsmühlen, J. Sehrt, and G. Witt, "Error detection in laser beam melting systems by high resolution imaging," in Proceedings of the Solid Freeform Fabrication Symposium, vol. 2012, 2012.
[15] H. Krauss, C. Eschey, and M. Zaeh, "Thermography for monitoring the selective laser melting process," in Proceedings of the Solid Freeform Fabrication Symposium, pp. 999-1014, 2012.
[16] R. A. Lyngby, J. Wilm, E. R. Eiríksson, J. B. Nielsen, J. N. Jensen, H. Aanæs and D. B. Pedersen, "In-line 3D print failure detection using computer vision," in Euspen and ASPE Special Interest Group Meeting: Additive Manufacturing, 2017.
[17] N. Perveen, D. Kumar, and I. Bhardwaj, "An overview on template matching methodologies and its applications," International Journal of Research in Computer and Communication Technology, vol. 2, no. 10, pp. 988-995, 2013.
[18] P. Patidar, M. Gupta, S. Srivastava, and A. K. Nagawat, "Image de-noising by various filters for different noise," International Journal of Computer Applications, vol. 9, no. 4, pp. 45-50, 2010.
[19] R. O. Julio, L. B. Soares, E. Costa, and S. Bampi, "Energy-efficient Gaussian filter for image processing using approximate adder circuits," in 2015 IEEE International Conference on Electronics, Circuits, and Systems (ICECS), pp. 450-453: IEEE, 2015.
[20] R. M. Haralick, S. R. Sternberg, and X. Zhuang, "Image analysis using mathematical morphology," IEEE Transactions on Pattern Analysis and Machine Intelligence, no. 4, pp. 532-550, 1987.
[21] R. C. Gonzales and R. E. Woods, "Digital image processing," ed: Prentice hall New Jersey, 2002.
[22] N. Otsu, "A threshold selection method from gray-level histograms," IEEE Transactions on Systems, Man, and Cybernetics, vol. 9, no. 1, pp. 62-66, 1979.
[23] Z. Qu and L. Zhang, "Research on image segmentation based on the improved Otsu algorithm," in 2010 Second International Conference on Intelligent Human-Machine Systems and Cybernetics, vol. 2, pp. 228-231: IEEE, 2010.
[24] S. Gupta and S. G. Mazumdar, "Sobel edge detection algorithm," International Journal of Computer Science and Management Research, vol. 2, no. 2, pp. 1578-1583, 2013.
[25] J. Canny, "A computational approach to edge detection," IEEE Transactions on Pattern Analysis and Machine Intelligence, no. 6, pp. 679-698, 1986.
[26] T. Smith Jr, W. Marks, G. Lange, W. Sheriff Jr, and E. Neale, "Edge detection in images using Marr-Hildreth filtering techniques," Journal of Neuroscience Methods, vol. 26, no. 1, pp. 75-81, 1988.
[27] J. Patel, J. Patwardhan, K. Sankhe, and R. Kumbhare, "Fuzzy inference based edge detection system using Sobel and Laplacian of Gaussian operators," in Proceedings of the International Conference & Workshop on Emerging Trends in Technology, pp. 694-697: ACM, 2011.
[28] T. Wuest, D. Weimer, C. Irgens, and K.-D. Thoben, "Machine learning in manufacturing: advantages, challenges, and applications," Production & Manufacturing Research, vol. 4, no. 1, pp. 23-45, 2016.
[29] V. Mnih, K. Kavukcuoglu, D. Silver, A. Graves, I. Antonoglou, D. Wierstra and M. Riedmiller, "Playing atari with deep reinforcement learning," arXiv preprint arXiv:1312.5602, 2013.
[30] C. J. Watkins and P. Dayan, "Q-learning," Machine learning, vol. 8, no. 3-4, pp. 279-292, 1992.
[31] V. Mnih1, K. Kavukcuoglu1, D. Silver1, A. A. Rusu1, J. Veness1, M. G. Bellemare1, A. Graves1, M. Riedmiller1, A. K. Fidjeland1, G. Ostrovski1, S. Petersen1, C. Beattie1, A. Sadik1, I. Antonoglou1, H. King1, D. Kumaran1, D. Wierstra1, S. Legg1 and D. Hassabis., "Human-level control through deep reinforcement learning," Nature, vol. 518, p. 529, 2015.
[32] U. Delli and S. Chang, "Automated process monitoring in 3D printing using supervised machine learning," Procedia Manufacturing, vol. 26, pp. 865-870, 2018.
[33] Y. Zhang, G. S. Hong, D. Ye, K. Zhu, and J. Y. Fuh, "Extraction and evaluation of melt pool, plume and spatter information for powder-bed fusion AM process monitoring," Materials & Design, vol. 156, pp. 458-469, 2018.
[34] IDS Imaging Development Systems GmbH official website. (December, 2020). Available: https://en.ids-imaging.com/store/ui-3240cp-rev-2.html.
[35] Computar official website. (December, 2020). Available: https://computar.com/product/601/MLH-10X.
[36] 梁靖偉，「彈性3D型成技術：鋅鋁合金與不銹鋼佈料塑形機構
與製程開發」，國立清華大學碩士論文，2019。