應用前饋神經網路預測CNC銑床之加工時間｜國立臺灣科技大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	陳孝行 Xiao-Xing Chen
論文名稱：	應用前饋神經網路預測CNC銑床之加工時間 Accurate Prediction of CNC Machining Time Using Feedforward Neural Networks
指導教授：	李維楨 Wei-Chen Lee
口試委員:	郭俊良 Chun-liang Kuo 劉孟昆 Meng-Kun Liu 王冬 Dong Wang 山本圭介 Keisuke Yamamoto
學位類別：	碩士 Master
系所名稱：	工程學院 - 機械工程系 Department of Mechanical Engineering
論文出版年：	2023
畢業學年度：	111
語文別：	英文
論文頁數：	65
中文關鍵詞：	CNC 、加工時間預測、類神經網路
外文關鍵詞：	CNC, machining time prediction, artificial neural network
相關次數：	點閱：228 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

電腦數值控制(Computer Numerical Control, CNC)加工機的加工時間預測對於工廠的生產力有著關鍵影響，當今商用電腦輔助製造(Computer-Aided Manufacturing, CAM)軟體通常具有加工時間預測功能，讓使用者在刀具路徑規劃時能夠預先得知零件的加工時間，然而由於商用CAM軟體並未考慮CNC加工機之機臺運動學以及機臺控制器之控制原理，使得加工時間的預測結果往往遠低於實際加工時間。為了解決上述問題，本研究發展出一套基於類神經網路的加工時間預測方法，研究首先利用電腦輔助設計(Computer-Aided Design, CAD)進行建模，再使用CAM產生出幾何形狀及曲面的NC程式，然後將其上傳至CNC加工機進行加工並實時擷取其加工資料，之後透過資料處理將機臺加工資料轉變為類神經網路的訓練資料。網路訓練的部分則是先建立一前饋神經網路(Feedforward Neural Network)，並且使用Levenberg-Marquardt作為訓練演算法，最後得到CNC加工機的加工時間預測模型。由於類神經網路是基於機臺加工資料進行訓練，使得其在考量機臺特性的前提下，相較於商用CAM軟體有更好的加工時間預測表現，實驗結果顯示，使用本研究所提出之加工時間預測模型能夠將預測誤差控制在2%以內，而商用CAM軟體則有約12%的預測誤差，這驗證了使用本研究所提出之方法預測機臺加工時間的可行性。

Prediction of Computer Numerical Control (CNC) machining time has a critical impact on productivity. Modern commercial Computer-Aided Manufacturing (CAM) software typically has machining time prediction capabilities, allowing users to anticipate the machining time for parts during toolpath planning. However, because commercial CAM software does not consider the actual CNC machine’s kinematics and the control principle of the CNC controller, the predicted machining time is often much shorter than the actual one. To address this problem, this study developed a neural network-based machining time prediction method. The study first used Computer-Aided Design (CAD) to model the part, then used CAM to generate the NC program. The NC program was then uploaded to the CNC machine for machining, and the machining data was captured in real-time. The machining data were then transformed into training data for the neural network through data pre-processing. The network training process involved building a feedforward neural network and using Levenberg-Marquardt as the training algorithm, resulting in a CNC machining time prediction model. As the neural network is trained based on machining data, it has better machining time prediction performance compared to commercial CAM software. Experimental results show that using the machining time prediction model proposed in this study can control prediction errors in 2%, while commercial CAM software has about 12% prediction error. This confirms the feasibility of using the proposed method to predict the machining time of CNC machines.

Abstract    I
摘要    II
致謝    III
Table of Contents    IV
List of Figures    VI
List of Tables    IX
Chapter 1    Introduction    1
1    Research Background    1
2    Literature Review    1
3    Objective    10
Chapter 2    Artificial Neural Network    11
1    Activation Function    12
2    Feedforward Neural Network    13
3    Gradient Descent    14
4    Backpropagation    14
5    Levenberg-Marquardt    15
Chapter 3    Materials and Methods    16
1    Materials and Setup    16
1.1    Tongtai CT-350 Machining Center    16
1.2    Siemens NX12    17
1.3    Visual C# and Traeger GmbH SINUMERIK .NET SDK    17
1.4    MATLAB    17
1.5    TensorFlow    17
2    Experimental Process    18
3    CAD Modeling and CAM Programming    18
3.1    CAD/CAM for Training Data    19
3.2    CAD/CAM for Model Performance Evaluation Part    20
4    Machining Data Acquisition    21
5    Training Data Pre-Processing    22
6    Neural Network Training    27
6.1    Introduction of MATLAB Neural Net Fitting App    27
6.2    Neural Network Setup    28
Chapter 4    Results and Discussion    29
1    MATLAB and TensorFlow Neural Network Training    29
2    Validation of prediction models using simple toolpaths    31
2.1    Short paths along the X axis    31
2.2    Long paths along the X axis    32
2.3    Arcs on the XY plane    33
2.4    Rectangles on the XY plane    36
2.5    Prediction results for complex planar toolpath    37
3    Prediction results for Benz mold    39
3.1    MATLAB and TensorFlow Model for Benz Mold Machining Time Prediction    40
4    The Inaccurate Prediction Made by NX    45
5    The Inaccurate Prediction Made by CNC Controller    47
Chapter 5    Conclusions and Future Work    49
1    Conclusions    49
2    Future Work    49
References    51


                                

[1] K. Saito, H. Aoyama, and N. Sano, "Accurate Estimation of Cutting Time Based on Control Principle of Machine Tool," International Journal of Automation Technology, vol. 10, no. 3, pp. 429-437, 2016, doi: 10.20965/ijat.2016.p0429.
[2] Y. Altintas and S. Tulsyan, "Prediction of part machining cycle times via virtual CNC," CIRP Annals, vol. 64, no. 1, pp. 361-364, 2015, doi: 10.1016/j.cirp.2015.04.100.
[3] C. Liu, Y. Li, W. Wang, and W. Shen, "A feature-based method for NC machining time estimation," Robotics and Computer-Integrated Manufacturing, vol. 29, no. 4, pp. 8-14, 2013, doi: 10.1016/j.rcim.2012.10.001.
[4] B. S. So, Y. H. Jung, J. W. Park, and D. W. Lee, "Five-axis machining time estimation algorithm based on machine characteristics," Journal of Materials Processing Technology, vol. 187-188, pp. 37-40, 2007, doi: 10.1016/j.jmatprotec.2006.11.159.
[5] E. Y. Heo, D. W. Kim, B. H. Kim, and F. Frank Chen, "Estimation of NC machining time using NC block distribution for sculptured surface machining," Robotics and Computer-Integrated Manufacturing, vol. 22, no. 5-6, pp. 437-446, 2006, doi: 10.1016/j.rcim.2005.12.008.
[6] R. T. Coelho, A. F. de Souza, A. R. Roger, A. M. Y. Rigatti, and A. A. de Lima Ribeiro, "Mechanistic approach to predict real machining time for milling free-form geometries applying high feed rate," The International Journal of Advanced Manufacturing Technology, vol. 46, no. 9-12, pp. 1103-1111, 2009, doi: 10.1007/s00170-009-2183-8.
[7] G. Medina-Sánchez, E. Torres-Jimenez, R. Lopez-Garcia, and R. Dorado-Vicente, "Cutting time in pocket machining for different tool-path approximation segments," Procedia Manufacturing, vol. 13, pp. 59-66, 2017, doi: 10.1016/j.promfg.2017.09.009.
[8] A. F. Souza, L. G. Camargo, H. S. Gaspar, F. Marin, A. C. Ochoa, and L. N. L. Lacalle, "New mechanistic model to predict machining time for milling free form geometries using 4-axis milling," Procedia CIRP, vol. 101, pp. 34-37, 2021, doi: https://doi.org/10.1016/j.procir.2020.09.187.
[9] H. Siller, C. A. Rodriguez, and H. Ahuett, "Cycle time prediction in high-speed milling operations for sculptured surface finishing," Journal of Materials Processing Technology, vol. 174, no. 1-3, pp. 355-362, 2006, doi: 10.1016/j.jmatprotec.2006.02.008.
[10] G. Komineas, P. Foteinopoulos, A. Papacharalampopoulos, and P. Stavropoulos, "Build Time Estimation Models in Thermal Extrusion Additive Manufacturing Processes," Procedia Manufacturing, vol. 21, pp. 647-654, 2018, doi: https://doi.org/10.1016/j.promfg.2018.02.167.
[11] H. Takizawa, H. Aoyama, and S. C. Won, "Basic Study on Rapid Estimation of Machining Time Based on AI with Two-Dimensional Data (Trihedral Figures)," International Journal of Materials, Mechanics and Manufacturing, vol. 7, no. 6, pp. 235-239, 2019, doi: 10.18178/ijmmm.2019.7.6.466.
[12] H. Takizawa, H. Aoyama, and S. C. Won, "Prompt Estimation of Die and Mold Machining Time by AI Without NC Program," International Journal of Automation Technology, vol. 15, no. 3, pp. 350-358, 2021, doi: 10.20965/ijat.2021.p0350.
[13] C. Sun, J. Dominguez-Caballero, R. Ward, S. Ayvar-Soberanis, and D. Curtis, "Machining cycle time prediction: Data-driven modelling of machine tool feed rate behavior with neural networks," Robotics and Computer-Integrated Manufacturing, vol. 75, 2022, doi: 10.1016/j.rcim.2021.102293.
[14] M. Endo and B. Sencer, "Accurate prediction of machining cycle times by data-driven modelling of NC system's interpolation dynamics," CIRP Annals, vol. 71, no. 1, pp. 405-408, 2022, doi: 10.1016/j.cirp.2022.04.017.
[15] A. Rodrigues, F. J. G. Silva, V. F. C. Sousa, A. G. Pinto, L. P. Ferreira, and T. Pereira, "Using an Artificial Neural Network Approach to Predict Machining Time," Metals, vol. 12, no. 10, 2022, doi: 10.3390/met12101709.
[16] Y. Yamamoto, H. Aoyama, and N. Sano, "Development of accurate estimation method of machining time in consideration of characteristics of machine tool," Journal of Advanced Mechanical Design, Systems, and Manufacturing, vol. 11, no. 4, pp. JAMDSM0049-JAMDSM0049, 2017, doi: 10.1299/jamdsm.2017jamdsm0049.
[17] J. G. Parmar, K. G. Dave, A. V. Gohil, and H. S. Trivedi, "Prediction of end milling process parameters using artificial neural network," Materials Today: Proceedings, vol. 38, pp. 3168-3176, 2021, doi: 10.1016/j.matpr.2020.09.644.
[18] G. Kalra and A. Kumar Gupta, "Mathematical modeling to estimate machining time during milling of Inconel 718 workpiece using ANN," Materials Today: Proceedings, 2022, doi: 10.1016/j.matpr.2022.11.314.
[19] T. Gurgenc, F. Ucar, D. Korkmaz, C. Ozel, and Y. Ortac, "A study on the extreme learning machine based prediction of machining times of the cycloidal gears in CNC milling machines," Production Engineering, vol. 13, no. 6, pp. 635-647, 2019, doi: 10.1007/s11740-019-00923-1.
[20] L. Li, F. Liu, B. Chen, and C. B. Li, "Multi-objective optimization of cutting parameters in sculptured parts machining based on neural network," Journal of Intelligent Manufacturing, vol. 26, no. 5, pp. 891-898, 2013, doi: 10.1007/s10845-013-0809-z.
[21] S. Samarasinghe, “Neural Networks for Applied Sciences and Engineering,” Auerbach Publications, 2007, doi: 10.1201/9780849333750
[22] 蘇木春、張孝德，「機器學習: 類神經網路、模糊系統以及基因演算法則」，全華圖書股份有限公司，2016，ISBN：9789864632060
[23] J. Fang, B. Li, H. Zhang, and P. Ye, "Real-time smooth trajectory generation for 3-axis blending tool-paths based on FIR filtering," The International Journal of Advanced Manufacturing Technology, vol. 126, no. 7, pp. 3401-3416, 2023, doi: 10.1007/s00170-023-11308-1.

全文公開日期本全文未授權公開 (校外網路)
全文公開日期本全文未授權公開 (國家圖書館：臺灣博碩士論文系統)

簡易檢索 / 詳目顯示

相關論文