臺灣博碩士論文加值系統

此項研究起源於印刷電路板製程中的缺陷。藉由先豐通訊公司提供的460天的數據資料，我們能藉此探討化銅製程中所須的各項藥液濃度異常和變化量異常與孔破發生的關係，而能建構出一個時間稽延及前饋類神經網路的整合系統用以預測孔破的發生的孔破預警系統。為了使預測最佳化，文中使用相關係數和均方根誤差來決定神經元和隱藏層的數目。在最佳化後，時間稽延類神經網路的預測結果和真實輸出的相關係數從0.3 提升至0.63。而每項藥液數據在訓練前皆經由前處理變成異常濃度和異常變化量兩種因子；從中也能定制出更為精準的藥液控制範圍。從而希望能達成藉由控制濃度而預防孔破發生的目標。在離線測試中，此系統能預測未成三天內孔破發生的機率，在第一天的準確率為0.96。

This study is originated due to the defects happening in printed circuit board manufacturing process. The study investigated the relationship between concentration of chemical and wall holes when the concentration is especially abnormal from the average. In this study a wall hole forecasting system is developed by combining time-delay and feedforward neural networks. In order to optimize the system, the method to decide numbers of hidden layers and hidden nodes in neural network has been investigated via two parameters of correlation coefficient and root mean square error. After optimization, the correlation coefficient compared between result nonlinear mapped by neural network and real data raised from 0.3 to 0.63. The system enables to predict the possibility for the wall hole happening in the next 3 days. In offline testing, the accuracy between real and predicted result is 0.96 for day one.

Chapter I: Introduction
1.1 Literature Survey……………………………………………………………..............1
1.2 Printed circuit board manufacturing process………………………… 4
Chapter II: Methodology
2.1 Preprocessing….…………………………………………………………............…6
2.2 Artificial neural network. …………………………………………………...13
2.3 Time-delay neural network…………………………………………….…...….14
2.4 Neuron analysis of depth of anesthesia case………..…21
2.5 The architecture of feed-forward neural network….....22
2.6 A hybrid system combined of time-delay and a feed-forward ANN………….........................................23
Chapter III: Results
3.1 TDNN...……………………………………………………………………................25
3.2 FFNN…………………………………………………………………….................. 28
3.3 Hybrid system combined by TDNN and FFNN…………………………29
Chapter IV Summary and Future Work……………………………………………….....33
References…………………………………………………………………………...................35

[1] Liu, C., Hu, W., & Sun, Y. (2013, July). Sensitivity analysis of PCB's design parameters of the plated through hole. In Quality, Reliability, Risk, Maintenance, and Safety Engineering (QR2MSE), 2013 International Conference on (pp. 489-492). IEEE.
[2] Su, F., Mao, R., Xiong, J., Zhou, K., Zhang, Z., Shao, J., & Xie, C. (2012). On thermo-mechanical reliability of plated-through-hole (PTH). Microelectronics Reliability, 52(6), 1189-1196.
[3] Huang, S. Q., Yung, K. C., & Sun, B. (2012). A finite element model and experimental analysis of PTH reliability in rigid-flex printed circuits using the Taguchi method. International Journal of Fatigue, 40, 84-96.
[4] Huang, X., Zhu, S., Huang, X., Su, B., Ou, C., & Zhou, W. (2015, August). Detection of plated through hole defects in printed circuit board with X-ray. In Electronic Packaging Technology (ICEPT), 2015 16th International Conference on (pp. 1296-1301). IEEE.
[5] Kurzmanowski, D., & Wilde, J. (2014, May). Using layout data for predicting effects on production processes in PCB-manufacturing. In Electronics Technology (ISSE), Proceedings of the 2014 37th International Spring Seminar on (pp. 315-320). IEEE.
[6] Moganti, M., Ercal, F., Dagli, C. H., & Tsunekawa, S. (1996). Automatic PCB inspection algorithms: a survey. Computer vision and image understanding,63(2), 287-313.
[7] Singh, A., Nayak, V. H., & Vayada, M. G. (2014). Automatic Detection of PCB Defects. International Journal for Innovative Research in Science and Technology, 1(6), 285-289.
[8] Ibrahim, I., Bakar, S., Mokji, M., Mukred, J., Yusof, Z. M., Ibrahim, Z. & Mohamad, M. S. (2012). A printed circuit board inspection system with defect classification capability. International Journal of Innovative Management, Information & Production ISME International, 3(1), 82-87.
[9] Doudkin, A., & Inyutin, A. (2014). The Defect and Project Rules Inspection on PCB Layout Image. International Journal of Computing, 5(3), 107-111.
[10] Mar, N. S. S., Yarlagadda, P. K. D. V., & Fookes, C. (2011). Design and development of automatic visual inspection system for PCB manufacturing. Robotics and Computer-Integrated Manufacturing, 27(5), 949-962.
[11] Iwahori, Y., Kumar, D., Nakagawa, T., & Bhuyan, M. K. (2012). Improved defect classification of printed circuit board using SVM. In Intelligent Decision Technologies (pp. 355-363). Springer Berlin Heidelberg.
[12] Iwahori, Y., Futamura, K., & Adachi, Y. (2011). Discrimination of true defect and indefinite defect with visual inspection using SVM. In Knowledge-Based and Intelligent Information and Engineering Systems (pp. 117-125). Springer Berlin Heidelberg.
[13] Duan, G., Wang, H., Liu, Z., & Chen, Y. W. (2012). A machine learning-based framework for automatic visual inspection of microdrill bits in PCB production. Systems, Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions on, 42(6), 1679-1689.
[14] Roh, B., Yoon, C., Ryu, Y., & Oh, C. (2001). A neural network approach to defect classification on printed circuit boards. Journal-Japan Society for Precision Engineering, 67(10), 1621-1626.
[15] An, S. S., No, B. O., Yu, Y. G., & Jo, H. S. (1996). A neural network approach to defect classification on printed circuit boards. Journal of Institute of Control, Robotics and Systems, 2(4), 337-343.
[16] Mironov, K. V., & PONGRATZ, M. (2013). Applying neural networks for prediction of flying objects trajectory. ВЕСТНИК УФИМСКОГО ГОСУДАРСТВЕННОГО АВИАЦИОННОГО ТЕХНИЧЕСКОГО УНИВЕРСИТЕТА, 17(6).
[17] Goyal, S., & Goyal, G. K. (2012). Time-Delay artificial neural network computing models for predicting shelf life of processed cheese. BRAIN. Broad Research in Artificial Intelligence and Neuroscience, 3(1), 63-70.
[18] Pandian, A., & Ali, A. (2014). Performance Prediction of an Automotive Assembly Line Based on ARMA-ANN Modeling. International Journal of Applied Industrial Engineering (IJAIE), 2(2), 22-39.
[19] Marcus, R. A. (1957). On the Theory of Oxidation‐Reduction Reactions Involving Electron Transfer. II. Applications to Data on the Rates of Isotopic Exchange Reactions. The Journal of Chemical Physics, 26(4), 867-871.
[20] Waszczyszyn, Z., & Ziemiański, L. (2001). Neural networks in mechanics of structures and materials–new results and prospects of applications. Computers & Structures, 79(22), 2261-2276.
[21] Yao, H. M., Vuthaluru, H. B., Tade, M. O., & Djukanovic, D. (2005). Artificial neural network-based prediction of hydrogen content of coal in power station boilers. Fuel, 84(12), 1535-1542.
[22] Hu, H. Y., Lee, Y. C., Yen, T. M., & Tsai, C. H. (2009). Using BPNN and DEMATEL to modify importance–performance analysis model–A study of the computer industry. Expert Systems with Applications, 36(6), 9969-9979.
[23] Kajitani, Y., Hipel, K. W., & McLeod, A. I. (2005). Forecasting nonlinear time series with feed‐forward neural networks: a case study of Canadian lynx data. Journal of Forecasting, 24(2), 105-117.
[24] Hornik, K., Stinchcombe, M., & White, H. (1989). Multilayer feedforward networks are universal approximators. Neural networks, 2(5), 359-366.
[25] Sanger, T. D. (1989). Optimal unsupervised learning in a single-layer linear feedforward neural network. Neural networks, 2(6), 459-473.
[26] Van Der Smagt, P. P. (1994). Minimisation methods for training feedforward neural networks. Neural networks, 7(1), 1-11.
[27] Jain, A., Tompson, J., Andriluka, M., Taylor, G. W., & Bregler, C. (2013). Learning human pose estimation features with convolutional networks. arXiv preprint arXiv:1312.7302.
[28] Fukushima, K. (1980). Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biological cybernetics, 36(4), 193-202.
[29] Werbos, P. (1974). Beyond regression: New tools for prediction and analysis in the behavioral sciences.
[30] LeCun, Y., Boser, B., Denker, J. S., Henderson, D., Howard, R. E., Hubbard, W., & Jackel, L. D. (1989). Backpropagation applied to handwritten zip code recognition. Neural computation, 1(4), 541-551.