臺灣博碩士論文加值系統

紡織業中的纖維分為許多種類，依製造的方式分為天然纖維及化學纖維，其中化學纖維的製作過程是將纖維素經過化學藥品處理成液體後，再透過紡口板上的紡嘴壓抽成絲，而紡口板是製作化學纖維最重要的元件之一，因此，在長時間製造化學纖維時，可能會導致紡嘴阻塞而影響生產的化學纖維品質，所以需要對紡口板進行檢測，而傳統的檢測都是藉由人工的方式，可能會導致穩定性不佳以及人工成本過高，因此，開發一套智慧自動檢測系統以取代人工並提高檢測的穩定性。
本系統使用水平關節機器人搭配深度學習以及自動光學檢測來檢測化學纖維紡口板上之紡嘴的阻塞情況，不僅可以取代人工上下料且降低人工成本，並利用深度學習提高檢測系統的準確性及穩定性，使整體系統達到全自動化的檢測系統。本系統的架構以雙Basler ace工業相機利用GigE方式連接、雙工業遠心鏡頭以及水平關節機器人組成，透過工業相機將當前影像進行處理並藉由深度學習模型檢測阻塞情況。本論文為檢測紡口板上之紡嘴，其紡口板分為兩種，環形排列和直行排列。環形排列的紡口板上有72孔紡嘴半徑為90 μm，直行排列則是有96孔紡嘴半徑為85 μm，當取得影像時會先將影像分割成數個250 x 250像素的小圖片，並將分割後的影像進行深度學習的標註和深度學習的訓練，透過大量的圖片數據讓訓練時進行多次的疊代，其主要是提升檢測的精確度，訓練完成後可在0.8秒內檢測單孔紡嘴的阻塞情況。本論文使用的深度學習是採用全卷積神經網路進行訓練，且採用語義分割的方式將影像中阻塞的像素點進行阻塞分類，最後分別計算出每一紡嘴阻塞的總面積。

In the textile industry, the fibers can be divided into natural fibers and chemical fibers according to the manufacturing method. The production process of the chemical fiber is to process the cellulose into liquid by chemical treatment, and then press and pull it into silk through the spinning nozzle on the spinnerets. The spinning nozzle is one of the most important components in the production of chemical fibers. However, the spinning nozzle may be blocked after long-term usage and the quality of chemical fibers may be affected. It is necessary to inspect the spinnerets, and the traditional inspections are performed by manual operation, which may lead to poor stability and high labor costs. Therefore, an intelligent automatic inspection system is developed to replace manual operation and improve the stability of inspection.
This research integrates SCARA robot, deep learning and Automatic optical inspection to inspect the blocked spinnerets while the chemical fiber is fabricated. It can not only replace manual refueling, but also can reduce labor costs. Also, the deep learning is used to improve the accuracy and stability of the inspection system. The purpose of this research is making the system fully automated.
The structure of this system uses dual Basler ace industrial cameras connected by GigE. Both cameras combined with industrial lenses and SCARA robots. The images is taken and processed by an industrial cameras to make a deep learning model, and the blockage is inspected by a deep learning model. The main purpose of this research is to inspect the spinning nozzle blocking. The spinnerets are divided into two types in this paper. The first one is circular arrangement, and the second one is straight arrangement. The circularly arranged spinneret has 72-hole spinning nozzles with a radius of 90 μm. And, straight arrangement has 96-hole spinning nozzles with a radius of 85 μm. When image has been captured, it will be divided to several smaller images, which is 250 × 250 pixels each. After that, the segmented images will be provided for deep learning annotation and training. While training, multiple iterations would be carried out through a large amount of image data. The main purpose is to improve the accuracy of inspection. After the training is completed, the blocking of the single-hole spinning nozzle can be inspected within 0.8 seconds.
In this research, Conventional Neural Network is used for training, and semantic segmentation is used to classify the blocked pixels in the image. Then, the total area of each spinning nozzle blocked are calculated and shown on software.

摘要...i
Abstract...ii
誌謝...iv
目錄...v
表目錄...viii
圖目錄...x
第一章 緒論...1
1.1 前言...1
1.2 研究目的...3
1.3 國內外相關文獻...3
1.3.1 關節式機器人之應用...4
1.3.2 機器視覺之應用...14
1.3.3 深度學習結合機器視覺之應用...24
1.4 論文架構...37
第二章 系統介紹...38
2.1 硬體架構...38
2.2 系統架構...39
2.3 電控系統元件架構及介紹...40
2.3.1 運動控制軸卡...40
2.3.2 AC伺服馬達驅動器...41
2.3.3 AC伺服馬達...43
2.3.4 PoE網路擴充卡...44
2.3.5 機器人控制器...44
2.3.6 水平關節機器人...45
2.4 視覺系統元件架構及介紹...47
2.4.1 Arduino UNO板...47
2.4.2 PWM調光驅動器...48
2.4.3 高角度環形光...50
2.4.4 LED燈板...50
2.4.5 PoE網路擴充卡...52
2.4.6 灰階攝影機...52
2.4.7 低倍率與高倍率遠心鏡頭...53
2.5 硬體配線...54
2.5.1 檢測平台配線...54
2.5.2 水平關節機器人控制器配線...56
2.6 其他設備介紹...57
2.6.1 夾爪...57
2.6.2 圖形處理器...58
2.6.3 人機介面、機器人控制器軟體及影像處理軟體...59
第三章 水平關節機器人與檢測系統之建置...60
3.1 水平關節機器人坐標系及動作定義...60
3.2 電控檢測平台校正...61
3.3 機器視覺檢測與演算法...65
3.3.1 軟體自動對焦...65
3.3.2 深度學習...68
3.4 鏡頭校正與阻塞分類...75
第四章 實驗結果與討論...76
4.1 水平關機器人教導軟體 (DRAStudio_V1.00.07.41)...76
4.2 人機介面 (Human-Machine Interface, HMI)...80
4.2.1 XYZ三軸智慧自動檢測平台...80
4.2.2 紡嘴檢測結果顯示介面...83
4.3 深度學習模型參數之訓練結果...85
4.4 實驗流程與檢測結果...86
4.4.1 紡口板之紡嘴掃描及阻塞檢測...88
4.5 結果討論...92
4.6 不確定因素分析...93
4.6.1 電控系統誤差...93
4.6.2 視覺系統誤差...93
第五章 結論與未來展望...94
5.1 結論...94
5.2 未來展望...94
參考文獻...95
附錄一 環形排列紡口板樣本一傳統檢測重複檢測...101
附錄二 環形排列紡口板樣本二傳統檢測重複檢測...102
附錄三 環形排列紡口板樣本三傳統檢測重複檢測...103
附錄四 環形排列紡口板樣本四傳統檢測重複檢測...104
附錄五 環形排列紡口板樣本五傳統檢測重複檢測...105
附錄六 環形排列紡口板樣本六傳統檢測重複檢測...106
附錄七 直行排列紡口板樣本七傳統檢測重複檢測...107
附錄八 直行排列紡口板樣本八傳統檢測重複檢測...108
附錄九 直行排列紡口板樣本九傳統檢測重複檢測...109
附錄十 環形排列紡口板樣本一深度學習重複檢測...110
附錄十一 環形排列紡口板樣本二深度學習重複檢測...111
附錄十二 環形排列紡口板樣本三深度學習重複檢測...112
附錄十三 環形排列紡口板樣本四深度學習重複檢測...113
附錄十四 環形排列紡口板樣本五深度學習重複檢測...114
附錄十五 環形排列紡口板樣本六深度學習重複檢測...115
附錄十六 直行排列紡口板樣本七深度學習重複檢測...116
附錄十七 直行排列紡口板樣本八深度學習重複檢測...117
附錄十八 直行排列紡口板樣本九深度學習重複檢測...118
Extended Abstract...119

[1]Textile Network, https://textile-network.com/en
[2]Semanticscholar, https://www.semanticscholar.org
[3]David Everett Rumelhart, Geoffrey Everest Hinton, Ronald J. Williams. Learning representations by back-propagating errors. Nature. 8 October 1986, 323 (6088): 533–536
[4]Siwen Fang, Xinlong Huang, Heping Chen, Ning Xi (2016). "Dual-arm robot assembly system for 3C product based on vision guidance". 2016 IEEE International Conference on Robotics and Biomimetics (ROBIO). pp. 807-812.
[5]Feifei. Zhao, Hao Gu, Cheng Li, Chanjuan Chen. (2017). "Accuracy analysis for robotized assembly system". 2017 IEEE International Conference on Robotics and Biomimetics (ROBIO). pp. 1850-1855.
[6]Mohannad Farag, Abdul Nasir Abd Ghafar, Mohammed Hayyan Alsibai. (2019). "Grasping and Positioning Tasks for Selective Compliant Articulated Robotic Arm Using Object Detection and Localization: Preliminary Results". 2019 6th International Conference on Electrical and Electronics Engineering (ICEEE). pp. 284-288.
[7]Xiuan Zhang; Xiaohui Xie. (2019). "A New Robot Body Calibration Method of SCARA Based on Machine Vision". 2019 IEEE International Conference on Robotics and Biomimetics (ROBIO). pp. 3077-3081.
[8]ArturoRealyvásquez-Vargas, Karina Cecilia Arredondo-Soto, Jorge Luis García-Alcaraz, Bogart Yail Márquez-Lobato, Jesrael Cruz-García. (2019). "Introduction and configuration of a collaborative robot in an assembly task as a means to decrease occupational risks and increase efficiency in a manufacturing company". Robotics and Computer-Integrated Manufacturing. Volume 57, June 2019, Pages 315-328.
[9]Mohannad Farag, Abdul Nasir Abd Ghafar, Mohammed Hayyan ALSIBAI. (2019). "Real-Time Robotic Grasping and Localization Using Deep Learning-Based Object Detection Technique". 2019 IEEE International Conference on Automatic Control and Intelligent Systems (I2CACIS). pp. 139-144.
[10]Fengming Li, Qi Jiang, Sisi Zhang, Meng Wei, Rui Song. (2019). "Robot skill acquisition in assembly process using deep reinforcement learning". Neurocomputing. Volume 345, 14 June 2019, Pages 92-102.
[11]Xiao Ling, Yuanshen Zhao, Liang Gong, Chengliang Liu, Tao Wang. (2019). "Dual-arm cooperation and implementing for robotic harvesting tomato using binocular vision". Robotics and Autonomous Systems. Volume 114, April 2019, Pages 134-143.
[12]Ziman Zhang, Zaojun Fang, Hongyuan Lian, Chi Zhang, Guilin Yang. (2021). "Robotic Peg-in-Hole Assembly System Based on Vision and Fuzzy Control". 2021 33rd Chinese Control and Decision Conference (CCDC). pp.5835-5839.
[13]Yuanpeng Liu, Jun Zhou, Yida Li, Yuqi Zhang, Yongqiang He, Jun Wang. (2022). "A high-accuracy pose measurement system for robotic automated assembly in large-scale space". Measurement. Volume 188, January 2022, 110426.
[14]Yong-Ju Jeon, Doo-chul Choi, Jong Pil Yun, Changhyun Park, Sang Woo Kim. (2011). "Detection of scratch defects on slab surface". 2011 11th International Conference on Control, Automation and Systems. pp. 1274-1278.
[15]M. Senthikumar, V. Palanisamy, J. Jaya. (2014). "Metal surface defect detection using iterative thresholding technique". Second International Conference on Current Trends In Engineering and Technology - ICCTET 2014. pp. 561-564.
[16]Li Jinlong, Duan Fan, Luo Lin, Gao Xiaorong. (2016). "Wheel profile and tread surface defect detection based on phase measuring profilometry". 2016 18th International Wheelset Congress (IWC). pp. 71-75.
[17]Ning Li, Jianyu Zhao, Ping Jiang. (2017). "Fabric defects detection via visual attention mechanism". 2017 Chinese Automation Congress (CAC). pp. 2956-2960.
[18]Qutong Chang, Yong Zhang, Zhe Sun. (2019). "Research on Surface Defect Detection Algorithm of Ice-cream Bars Based on Clustering". 2019 IEEE 3rd Information Technology, Networking, Electronic and Automation Control Conference (ITNEC). pp. 537-541.
[19]Naijian Chen, Jianbo Sun, Xu Wang, Yulin Huang, Yingjun Li, Chao Guo. (2019). "Research on surface defect detection and grinding path planning of steel plate based on machine vision". 2019 14th IEEE Conference on Industrial Electronics and Applications (ICIEA). pp. 1748-1753.
[20]Xianen Zhou, Yaonan Wang, Qing Zhu, Jianxu Mao, Changyan Xiao, Xiao Lu, Hui Zhang. (2020). "A Surface Defect Detection Framework for Glass Bottle Bottom Using Visual Attention Model and Wavelet Transform". IEEE Transactions on Industrial Informatics. Volume: 16, Issue: 4, April 2020,
[21]Huailiang Zhang, Ling Peng, Sheng Yu, Wei Qu. (2021). "Detection of Surface Defects in Ceramic Tiles with Complex Texture". IEEE Access. Volume: 9.
[22]Fangyu Liu, Linbing Wang. (2022). "UNet-based model for crack detection integrating visual explanations". Construction and Building Materials. Volume 322, 7 March 2022, 126265.
[23]Yasser M. Fouda. (2020). "Integral images-based approach for fabric defect detection". Optics & Laser Technology. Volume 147, March 2022, 107608.
[24]Guanlin Liu. (2020). "Surface Defect Detection Methods Based on Deep Learning: a Brief Review". 2020 2nd International Conference on Information Technology and Computer Application (ITCA). pp. 200-203.
[25]Adriana Birlutiu, Adrian Burlacu, Manuella Kadar, Daniela Onita. (2017). "Defect Detection in Porcelain Industry Based on Deep Learning Techniques". 2017 19th International Symposium on Symbolic and Numeric Algorithms for Scientific Computing (SYNASC). pp. 263-270.
[26]Ruturaj Kulkarni, Shruti Dhavalikar, Sonal Bangar. (2018). "Traffic Light Detection and Recognition for Self Driving Cars Using Deep Learning". 2018 Fourth International Conference on Computing Communication Control and Automation (ICCUBEA). pp. 1-4.
[27]Kudrov Maksim, Bukharov Kirill, Zakharov Eduard, Grishin Nikita, Bazzaev Aleksandr, Lozhkina Arina, Semenkin Vladislav, Makhotkin Daniil. (2019). "Classification of Wafer Maps Defect Based on Deep Learning Methods with Small Amount of Data". 2019 International Conference on Engineering and Telecommunication (EnT). pp. 1-5.
[28]Bing Wei, Kuangrong Hao, Lei Gao, Xue-songTang. (2020). "Detecting textile micro-defects: A novel and efficient method based on visual gain mechanism". Information Sciences. Volume 541, December 2020, Pages 60-74.
[29]Shancheng Tang, Xiongxiong Chen, Puyue Zhang, Ming Chen, Hanbo Wang. (2020). "Appearance Detection Method of Resistance Based on Deep Learning". 2020 International Conference on Intelligent Transportation, Big Data & Smart City (ICITBS). pp. 856-859.
[30]Zhenyu Liu, Ruining Tang, Guifang Duan, Jianrong Tan. (2021). "TruingDet: Towards high-quality visual automatic defect inspection for mental surface". Optics and Lasers in Engineering. Volume 138, March 2021, 106423.
[31]Feng Xu, Yajun Liu, Bin Zi, Lei Zheng. (2021). "Application of Deep Learning for Defect Detection of Paint Film". 2021 6th International Conference on Intelligent Computing and Signal Processing (ICSP). pp. 1118-1121.
[32]Jiahui Yao, Junfeng Li. (2022). "AYOLOv3-Tiny: An improved convolutional neural network architecture for real-time defect detection of PAD light guide plates". Computers in Industry. Volume 136, April 2022, 103588.
[33]Zhuxi MA, Yibo Li, Minghui Huang, Qianbin Huang, Jie Cheng, Si Tang. (2022). "A lightweight detector based on attention mechanism for aluminum strip surface defect detection". Computers in Industry. Volume 136, April 2022, 103585.
[34]林伯軒，應用深度學習與機器視覺於化學纖維紡嘴智慧自動檢測系統開發，碩士論文，國立虎尾科技大學自動化工程所碩士班，台灣，2020。
[35]Delta, http://www.delta.com.tw
[36]ADLINK Technology, https://www.adlinktech.com
[37]Arduino, https://www.arduino.cc
[38]MEAN WELL, https://www.meanwell.com
[39]Chen Hsuan, https://www.chtco.com.tw
[40]Basler, https://www.baslerweb.com
[41]Myutron, https://www.myutron.com
[42]Auspicious Automation, http://airbank.com.tw
[43]ASUS, https://www.asus.com/tw
[44]Keysight, https://www.keysight.com
[45]影像處理與電腦視覺-鍾國亮(第4.2章節102頁)
[46]https://ithelp.ithome.com.tw/
[47]https://blog.csdn.net/jacke121
[48]https://www.twblogs.net/a/5d6687c2bd9eee5327feb739