臺灣博碩士論文加值系統

隨著電子產品的薄型化發展趨勢，微孔板廣泛地應用在手機板以及高階的IC封裝載板。在微成孔的製程裡，雷射鑽孔由於可提供彈性、高速且精確的微孔加工，近年來已逐漸成為主流技術。
雷射鑽孔最常發生的缺陷就是縮孔現象，其形成的原因有很多，諸如雷射功率、脈波寬度、脈波頻率、脈波數目、輔助氣體壓力以及聚焦深度…等等。
由於機台的調校參數過於複雜而且彼此間又互相影響，不容易以一般的物理定律進行推導，本研究使用類神經網路將機台的複雜行為予以模式化並分別規劃雷射鑽孔、雷射切孔與雷射切軸三個實驗來探討類神經網路對於不同加工製程的應用問題，最後建立一套能夠計算製程參數最適值的數學系統，希望能夠藉此幫助製程相關人員縮短調機時間以及提升加工品質。

Modern electrical products display itself with the feature of not only versatile but also miniature. The request of the product creation force to develop more complicated integrated circuits and multiple layers Printed Circuit Board, PCB. Laser micro drilling technology can provide flexible, high speed, precise manufacturing process has become the most popular drilling technology, especially in small hole (< 1 mm diameter) drilling of electrical industry.
But laser micro drilling technology also has some manufacturing problems such as hole tapering, barreling, surface debris and nonuniform hole diameter. These defects are the results of unsuitable drilling parameters. How to search for the correct operating parameters and then improving the qualities is the aim of the research.
Generally, laser micro drilling process has about seven parameters to be properly adjusted, and the correlations between these parameters are nonlinear and complicated. It is difficult for a fresh engineer to be familiar with the operation in a short time. This paper reports an investigation of the quality control technique of laser micro drilling process.
First, using the One-Factor-at-a-Time experiments, we can decide the levels of parameters. Second, adopting the Full Factorial experiment, we can get the training sets of neural network. Third, using training sets and neural network to find the mathematical model between parameters and machining qualities. Finally, by the calculation of neural network, we can find the optimal parameters of the laser micro drilling process. The result of the study can assist to generate the feasible parameters of laser micro drilling machine and improve the quality of the product.

目 錄 i
圖目錄 iii
表目錄 v
摘要 1
Abstract 2
誌謝 3
第一章、緒論 4
1.1 研究動機 4
1.2 文獻回顧 6
1.3 研究目的 7
1.4 實驗流程 8
1.5 論文架構 9
第二章、雷射加工技術 11
2.1 雷射發展簡史 11
2.2 雷射的特性 12
2.3 雷射加工技術基本理論 13
2.4 雷射鑽孔技術 15
第三章、研究方法 18
3.1 田口實驗計劃法 18
3.1.1 損失函數 18
3.1.2 品質特性的種類 20
3.1.3 傳統實驗方法與田口實驗計劃法 22
3.1.4 直交表 23
3.1.5 訊號雜音（S/N）比 24
3.2 類神經網路 26
3.2.1 類神經網路發展簡史 26
3.2.2 生物神經元模型 28
3.2.3 人工神經元模型 29
3.2.4 類神經網路的組成架構 31
3.2.5 類神經網路的學習 34
第四章、實驗流程與設備 36
4.1 實驗流程 37
4.2 實驗機台介紹 41
4.3 實驗項目介紹 43
4.4 加工品質評定 49
4.5 類神經網路重要參數設定 59
第五章、實驗結果 61
5.1 實驗環境設定 61
5.2 雷射鑽孔實驗 61
5.3 雷射切孔實驗 74
5.4 雷射切軸實驗 86
5.5 類神經網路模擬程式 97
第六章、結論與建議 98
6.1 結論 98
6.2 後續研究方向 99
參考文獻 100
附錄一、 L27（313）直交表 103
附錄二、雷射鑽孔加工參數表 104
附錄三、雷射切孔加工參數表 107
附錄四、雷射切軸加工參數表 109

[ 1] L.Li, D.K.Y. Low and M.Ghoreshi, “Hole Taper Characterisation and Control in Laser Percussion Drilling”, Laser Processing Research Centre, UMIST, Manchester, UK
[ 2] Yilbas, B.-S., 1987, “Study of Affecting Parameters in Laser Hole Drilling of Sheet Metals”, Trans. ASME: J. Eng. Mat.&Tech., 109: 282-287
[ 3] Hamoudi, W.-K., Rasheed, B.-G., 1995, “Parameters Affecting Nd:YAG Laser Drilling of Metals”, International Journal for the Joining of Materials, 7:63-69
[ 4] Yilbas, B.-S. and Yilbas, Z., “Parameters affecting hole geometry in laser drilling of Nimonic 75”, Lasers in Motion for Industrial Applications, Proceedings of the SPIE, Vol. 744, 1987, pp. 87-91
[ 5] K. Hornik, “Multilayer Feedforward Networks are Universal
Approximators”, Neural Networks, Vol. 2, pp. 359-366, 1989
[ 6] U. Bhattacharya, “Self-adaptive Learning Rates in Backpropagation Algorithm Improve Its Function Approximation Performance”, 1995. Proceedings., IEEE International Conference on Neural Networks, Vol. 5, pp. 2784-2788, 1995
[ 7] Yo-Ping Huang, “Improving the Backpropagation Learning Speed with Adpative Neuro-Fuzzy Technique”, Proceeding of 1993
International Joint Conference on Neural Networks, Vol 3,
pp. 2897-2900, 1993
[ 8] A. Zaghw, “An Automated Approach for Selecting the Learning Rate and Momentum in Back-Propagation Networks”, IEEE International Conference on Neural Networks, Vol 1, pp. 464-469, 1994
[ 9] M.E. Rimer, “Speed Training: Improving the Rate of
Backpropagation Learning through Stochastic Sample Presentation”, IEEE 2001
[10] D. Nguyen, “Improving the Learning Speed of 2-Layer Neural
Networks by Choosing Initial Values of the Adaptive Weights”,
IJCNN International Joint Conference on Neural
Networks,vol.3 ,pp.21-26,1990
[11] Yolanda M. Pirez and Dilip Sarkar, ”Back-Propagation Algorithm with Controlled Oscillation of Weights”, Neural Networks, IEEE International Conference 1993, vol.1, P21-26
[12] Hagan,“Neural Network Design”, PWS PUBLISHING COMPANY
1995
[13] S.Haykin, “Neural Networks : A Comprehensive Foundation”,
Prentice Hall 1999
[14] N. Sundararajan, “Parallel Architectures for Artificial Neural Networks”, IEEE Computer Society, 1998
[15] C. Principe, “Neural and Adaptive Systems”, John Wiley & Sons Inc. 1999
[16] F.Scradelli, A.H. Tsoi, “Universal Approximation Using Feedforward Neural Networks: A Survey of Some Existing Methods, and Some New Results”, Neural Networks, 1998
[17] M.Riedmiller, “A Director Adaptive Method for Faster
Backpropagation Learning: The RPROP Algorithm”, IEEE 1993
[18] L. Marquez, “Function Approximation Using Backpropagation and General Regression Neural Networks”, IEEE 1993
[19] David S. Chen, “A Robust Back Propagation Learning Algorithm for Function Approximation”, IEEE 1994
[20] Martin T. Hagan, “Training Feedforward Networks with the
Marquart Algorithm”, IEEE 1994
[21] J. Wang, “On the Design Principles of the Functional Link Nets”, IEEE 1990
[22] W.J. Kim, “Application of Neural Networks to Signal Prediction in Nuclear Power Plant”, IEEE 1993
[23] J. Lee, “Improvement on Function Approximation Capability of Backpropagation Neural Networks”, 1991 IEEE International Joint Conference on Neural Networks, vol.2, pp.1367-1372,1991
[24] 謝德宇，「雷射切雕機的精度調校自動化」，國立交通大學機械
工程研究所, 2002
[25] 黃偉治，「以類神經網路控制可變阻尼及勁度結構之位移研究」，
國立交通大學土木工程研究所, 2002
[26] 謝冠華，「類神經網路應用於機台設定及設備選擇: 以塑模機為
例」，國立清華大學工業工程研究所, 2000
[27] 羅華強，類神經網路-MATLAB的應用，清蔚科技, 2001
[28] 葉怡成，類神經網路模式應用與實作，儒林圖書, 1997
[29] 蔡宗河，CO2雷射加工，全華科技, 1995
[30] 林三寶，雷射原理與應用，全華科技, 1987