臺灣博碩士論文加值系統

電解式線上削銳(Electrolytic In-process Dressing, ELID)常用於硬脆材料上，可得良好的工件精度與粗糙度，然而其運用在金屬等延性材料上，卻容易因加工過程中易發生切屑填塞現象，而造成工件表面粗糙度不佳，為改善此種現象，本研究加入壓電制動元件(PIEZO)，利用其高速往復微小振動的特性作用於工件上，使工件在加工過程中產生微小的位移量，延長砂輪與工件接觸的距離，而且能幫助砂輪的排屑，減少切屑填塞的可能，並與未加壓電制動元件的實驗作比較，了解粗糙度改善的程度。
類神經網路在實驗建模上已應用多時，其中放射狀基底函數網路(Radial-Basis Function Network)為一套完整的理論，其最大特色為可將非線性轉成線性輸出，故本研究對其作深入之探討並利用其作為實驗建模的工具。最後再應用遺傳演算法之最佳化方法，計算出最佳之加工參數並驗證之。

Electrolytic In-process Dressing (ELID) was developed mainly to machine brittle materials and it can achieve lower surface roughness and higher machining rate. However, while using ELID grinding to process metals, loading has become a severe problem. In order to overcome this problem, we use actuators PIEZO in the research to shake the working piece while in processing. Due to its quick micro oscillating motion, the PIEZO will help to extend the contact distance between the grinding wheel and the working piece. Furthermore, the loading problem is avoided due to the vibration motion. We then compare it with the traditional ELID grinding to check how the surface roughness is improved.
Artificial Neural Network has been used in building non-analytical process models, which often are highly non-linear system. Within ANN domain, the Radial-Basis Function Network has been developed based on some analytical mathematical theories, and we chose to use it for building the ELID grinding model. Finally, the Genetic Algorithms optimization method is used to obtain the optimal grinding process parameters.

第一章 緒論
1.1前言......................................................1
1.2文獻回顧..................................................2
1.3研究動機..................................................5
第二章 相關理論
2.1電解式線上削銳(ELID)......................................7
2.1.1原理 ....................................................7
2.1.2組成 ....................................................9
2.1.3電解式線上削銳與傳統電解削銳比較........................16
2.1.4研磨的特色..............................................18
2.2類神經網路(Artificial Neural Network).....................20
2.2.1活化函數................................................22
2.2.2類神經網路的學習規則....................................26
2.2.3代價函數................................................28
2.3倒傳遞類神經網路(backpropagation networks)................30
2.3.1倒傳遞網路架構..........................................31
2.3.2倒傳遞演算法則..........................................31
2.4 Radial-Basis Function Network (放射狀基底函數網路) ......35
2.4.1 COVER’S非線性問題之分類理論...........................35
2.4.1.1 Cover''s theorem......................................35
2.4.1.2 Separating Capacity of a Surface.....................41
2.4.2 INTERPOLATION PROBLEM ..................................42
2.4.2.1 Micchelli’s Theorem.................................45
2.4.3 SUPERVISED LEARNING AS AN ILL-POSED HYPER-
SURFACE RECONSTRU PROBLEM..............................45
2.4.3.1 REGULARIZATION THEORY................................47
2.4.3.2 Green’s Function....................................49
2.4.3.3 Regularization Problem之解...........................51
2.4.3.4 Expansion Coefficients之計算.........................52
2.4.3.5 Multivariate Gussian Functions.......................56
2.4.4 REGULARIZATION NETWORKS................................58
2.4.5 GENERALIZED RADIAL BASIS FUNCTION NETWORKS.............60
2.5 遺傳演算法...............................................64
第三章 實驗方法與結果
3.1實驗裝置..................................................69
3.2實驗規劃..................................................73
3.3實驗結果..................................................73
3.4 ELID研磨驗證實驗規劃表...................................76
第四章 實驗model之建立與分析
4.1研究方法..................................................77
4.2實驗model建立之程式流程圖.................................78
4.3實驗結果分析..............................................80
4.3.1粗糙度改善程度..........................................80
4.3.2無PIEZO之模式分析.......................................81
4.3.3加裝PIEZO之模式分析.....................................84
4.3.4有無PIEZO之比較.........................................87
4.4實驗結果與磨削條件關係....................................87
4.4.1無PIEZO.................................................88
4.4.2加裝PIEZO...............................................89
4.4.3與SAS分析結果做比較.....................................91
4.5切削力討論................................................92
4.6測試建立model之預測能力...................................93
第五章 實驗之最佳化與驗證
5.1最佳化之方法..............................................97
5.2放射狀基底函數網路(RBF) model之實驗參數最佳化.............99
5.3倒傳遞類神經網路(BP) model之實驗參數最佳化................100
第六章 結論與未來展望
6.1結論......................................................102
6.2未來展望..................................................103
參考文獻.....................................................105

[1]Ohmori H. and Nakagawa T., “Grinding of Silicon Using Cast Iron Fiber Bonded Wheel,” Preprint for Autum Conference of JSPE, p687, 1987.
[2]Ohmori H. and Nakagawa T., “Mirror Surface Grinding of Silicon Wafers with Electrolytic In-Process Dressing”, Annals of the CIRP, Vol.39, No.1, p329, 1990.
[3]Ohmori H., “Electrolytic In-Process Dressing (ELID) Grinding Technique for Ultra-precision Mirror Surface Machine”, Int.Journal JSPE, Vol.26, No.4, p.273, 1992.
[4]Ohmori H., "Efficient and Precision Grinding Technique for Ceramics with Electrolytic In-Process Dressing (ELID)", NIST Special Publication Proceedings of the International Conference on Machining of Advanced Materials, 1993.
[5]Ohmori H. and Nakagawa T., “Analysis of Mirror Surface Generation of Hard and Brittle Materials by ELID Grinding with Superfine Grain Metakkic Bond Wheels”, Annals of the CIRP, Vol.44, No.1, p287, 1995.
[6]Bandoypadhyay B.P., Ohmori H., and Takahashi, I., "Efficient and Stable Grind of Ceramics by Electrolytic In-Process Dressing (ELID)", Journal of Materials Processing Technology, Vol.66, n1-3, Apr.1997.
[7]Ohmori H. and Nakagawa T., “Utilization of Nonlinear Condition in Precision Grinding with ELID (Electrolytic In-Process Dressing) for Fabrication of Hard Material Components, ” Annals of the CIRP, Vol.46, No.1, p261, 1997.
[8]Itoh Nobuhide, Ohmori Hitoshi, and Bandoypadhyay B.P., "Grinding Characteristics of Hard and Brittle Materials by ELID-lap Grinding Using Fine Grain Wheels", Materials and Manufacturing Processes, Vol.12, NO.6, Nov, 1997.
[9]Bandoypadhyay B.P., Ohmori H., "The effect of ELID grinding on the flexural strength of silcon nitrude", Int. Machine tools and Manufacture, Vol.39, p839-853, 1999.
[10]Eun-Sang Lee, “A study on the mirror-like grinding of die steel with optimum in-process electrolytic dressing”,Journal of Materials Processing Technology ,Vol.100,p200-208,2000.
[11]D.J.Stephenson*,D.Veselovac,S.Manley, J.Corbett,“Ultra-precision grinding of hard steels”,Precision Engineering, Journal of the International Socities for Precision Engineering and Nanotechnology, Vol,25,p336-345,2001.
[12]黃銘銓,"電解式線上削銳系統磨削性能與機構之研究"，碩士論文，國立台灣海洋大學，2001.
[13]曹明綜,"RBF類神經網路與遺傳演算法在精密磨拋加工建模及最佳化之應用",碩士論文，國立台灣海洋大學，2001.
[14]中國砂輪研磨加工技術叢書編輯委員會，"工業升級的新伙伴，高效率、高品質、超精密研磨的鏡面加工技術─ELID研磨"，中國砂輪企業股份有限公司。
[15]劉應進"精密陶瓷電解式線上削銳之研究"，碩士論文，國立台灣海洋大學，2000。
[16]蘇木春、張孝德著，”機器學習類神經網路、模糊系統以及基因
演算法則”，全華科技圖書股份有限公司，1997。
[17]Simon Haykin, “Neural Networks: A comprehensive Foundation”, 2nd edition Prentice-Hall,1999.
[18]Cover, T. M., “Geometrical and statistical properties of system of linear inequalities with applications in pattern recognition,” IEEE Transactions on Electronic Computers, vol. EC-14, pp.326-334, 1965.
[19]Mhaskar, H.N., “Neural networks for optimal approximation of smooth and analytic functions,” Neural Computation, vol.8, pp.1731-1742,1996.
[20]Niyogi, P., and F. Girosi, 1996. “On the relationship between generalization error, hypothesis complexity, and sample complexity for radial basis functions,” Neural Computation, vol.8, pp.819-842.
[21]Broomhead, D.S., and D. Lowe, 1988. “Multivariable function interpolation and adaptive networks,” Complex Systems, vol.2, pp.321-355.
[22]Micchelli, C.A., 1986. “Interpolation of scattered data: Distance matrices and conditionally positive definite functions,” Constructive Approximation, vol.2, pp.11-22.
[23]Tikhonov, A.N., and V.Y.Arsenin, 1997. Solutions of Ill-posed Problems, Washington, DC: W.H.Winston.
[24]Morozov, V.A., 1993. Regularization Methods for Ill-Posed Problems, Boca Raton, FL: CRC Press.
[25]Kirsch, A., 1996. An Introduction to the Mathematical Theory of Inverse Problems, New York: Springer-Verlag.
[26]Tikhonov, A.N., 1963. “On solving incorrectly posed problems and method of regularization,” Doklady Akademii Nauk USSR, vol.151 pp.501-504.
[27]Dorny, C.N., 1975. A Vector Space Approach to Models and Optimization, New York: Wiley (Interscience).
[28]Debnath,L., and P. Mikusinki, 1990. Introduction to Hilbert Space with Applications, New York: Academic Press.
[29]Do Figueiredo, R.J.P., and G. Chen, 1993. Nonlinear Feedback Control Systems, New York: Academic Press.
[30]Lancoz, C., 1964. Linear Differential Operators, London: Van Nostrand.
[31]Courant, R., and D. Hilbert, 1970. Methods of Mathematical Physics, vol.ⅠandⅡ, New York: Wiley Interscience.
[32]Poggio, T., and F. Girosi, 1990a. “Networks for approximation and learning,” Proceedings of the IEEE, vol.78, pp.1481-1497.
[33]錢明淦著，遺傳演算法應用於具有多種資源組態及資源限制專案計劃排程問題之研究 ，元智大學工業工程研究所碩士論文，1999。
[34]D.B. Fogel, An Evolutionary Approach to the Traveling Salesman Problem, Biological Cybernetics, vol.60, pp.139-144, 1988.
[35]D.E. Goldberg, Genetic Algorithms: in Search Optimization and Machine Learning, Reading, MA: Addison-Wesley, 1989.
[36]Kenji Yamamoto and Osamu Inoue, New Evolutionary Direction Operator for Genetic Algorithms, AIAA Journal, vol.33, No.10, pp.1990-1993, 1995.
[37]K.F. Man, K.S. Tang, and S.K. Wong, Genetic Algorithms: Concepts and Applications, IEEE Transactions On Industrial Electronics, Vol.43, No.5, pp.519-533, 1996.
[38]A. Scott, An Introduction to Genetic Algorithms, AI Expert, Vol.4, No.3, p.49-53, 1990.
[39]M. Srinivas and L.M. Patnaik, Genetic Algorithms: A Survey, IEEE Computer, June, pp.17-26, 1994.
[40]S.Malkin,”GRINDING TECHNOLOGY : Theory and Application of Machining with Abrasive”,pp.181,1989.