隨著微細加工技術的進步，生物晶片尺寸也隨之日趨縮小，雖然可大幅地提升儀器的攜帶性，但是晶片的空間也因此受到限制，不利於毛細管電泳的分析，毛細管勢必要藉由多次轉彎來獲得足夠長度，使不同的物質充份分離，不過轉彎的管道將導致樣本偏斜量(sampling skewness)增加造成偵測困難，因此要如何降低偏斜量將是本文所探討的重點。
本文使用「類神經網路結合複合型法」對彎曲管道進行最佳化設計，以降低偏斜量。首先定義毛細管電泳的設計參數，訂出各設計參數的水準值，再利用田口式直角表進行參數配置，以獲得參數均勻配置的範例。之後，以計算流體力學(CFD)方法來模擬不同設計參數的電泳現象，並計算獲得偏斜量，再利用類神經網路的學習能力建構設計參數與偏斜量之間的關係，最後將類神經網路結合複合型法並進行最佳化搜尋，找出最佳設計參數，以提供設計者一個正確、快速的設計途徑。

With improving the micro-machining technology, the size of the bio-MEMS is reduced gradually. The reduced-size chips make the apparatus more portable. On the other hand, because of the limited space, the capillary electrophoresis has to serpentine to get sufficient length. It needs enough length to make the different materials to be separated. But the serpentine channel results in the larger sampling skewness, which makes the detection to become more difficult. And how to reduce the sampling skewness is the main consideration in this study.
The approach combining the artificial neural network with the complex method is used to optimize the serpentine channel for reducing the sampling skewness. First, the design parameters that construct the capillary electrophoresis are defined. Five levels are taken for each design parameters. The orthogonal array of Taguchi method is used to group up the design parameters. The sampling skewnesses of different cases are obtained from the CFD (computational fluid dynamics) simulations. The relationship between the design parameters and the sampling skewness is established by training an artificial neural network. Then the combination of the well-trained network with the complex method is used to search the optimum design parameters for the turning geometry. At last, it is hoped to provide the manufacturers the way to design the capillary electrophoresis.

[1] 傅龍明, ”微晶片電泳流場之分析與應用,” 國立成功大學工程科學研究所博士論文(2001).
[2] H. V. Helmholtz, Wiedemanns Ann. 7, 337 (1877).
[3] F. Kohlrausch, Wiedemanns Ann. 62, 209 (1897).
[4] Tiselius, A., Trans. Faraday. Soc., 33, 524 (1937).
[5] 連香媚, ”利用毛細管電泳分析DNA在臨床上的應用,” 國立成功大學化學研究所碩士論文(2000).
[6] Virtanen R., Acta Polytech. Scand., 123, 1-67 (1974).
[7] Jorgenson, J. W. and Lukacs, K. D., “Zone Electrophoresis in Open-Tubular Glass Capillaries,” Analytical Chemistry, 53, pp. 1298-1302(1981).
[8] Terabe, S., Otsuka, K., Ichikawa K., Tsuchiya, A. and Ando, T., “Electrokinetic Separations with Micellar Solutions and Open-Tubular Capillaries,” Analytical Chemistry, 56, pp. 111-113(1984).
[9] Hjerten, S., Liao. J. and Yao, K., J. Chromatographia, 387, 127(1987).
[10] Rose, D. J. and Jorgenson, J. W., “Characterization and Automation of Sample Introduction Methods for Capillary Zone Electrophoresis,” Analytical Chemistry, 60, pp. 642-648(1988).
[11] Harrison, D. J., Manz, A., Fan, Z., Ludi, H. and Widmer M. H., “Capillary Electrophoresis and Sample Injection Systems Integrated on a Planar Glass Chip,” Analytical Chemistry, 64, pp. 1926-1932(1992).
[12] Seiler, K., Harrison, D. J. and Manz, A., “Planar Glass Chips for Capillary Electrophoresis: Repetitive Sample Injection, Quantitation, Separation Efficiency,” Analytical Chemistry, 65, pp. 1481-1488(1993).
[13] Effenhauser, C. S. and Manz, A., “Glass Chip for High-Speed Capillary Electrophoresis Separations with Submicrometer Plate Heights,” Analytical Chemistry, 65, pp. 2637-2642 (1993).
[14] Jacobson, S. C., Hergenroden, R., Koutny L. B., Warmack R. J. and Ramsey J. M., “Effects of Injection Schemes and Column Geometry on the Performance of Microchip Electrophoresis Devices,” Analytical Chemistry, 66, pp. 1107-1113(1994).
[15] Culbertson, C. T. and Jacobson, S. C., “Dispersion Sources for Compact Geometries on Microchips,” Analytical Chemistry, 70, pp. 3781-3789 (1998).
[16] Paegel, B. M. and Mathies, R. A., “Turn Geometry for Minimizing Band Broadening in Microfabricated Capillary Electrophoresis Channels,” Analytical Chemistry, 72, pp. 3030-3037(2000).
[17] Molho, J. M., Herr, A. E., Mosier, B. P., Santiago, J. G., Kenny, T. W., Bernnen, R. A. and Gordon, G. B., “Designing Corner Compensation for Electrophoresis in Compact Geometries,” Micro Total Analysis Systems, Enschede, The Netherlands(2000).
[18] Grushka, E., McCormick, R. M. and Kirkland, J. J., “Effect of Temperature Gradients on the Efficiency of Capillary Zone Electrophoresis Separations,” Analytical Chemistry, 61, pp. 241-246 (1989).
[19] Griffiths, S. K. and Nilson, R. H., “ Band Spreading in Two-DimensionalMicrochannel Turns for Electrokinetic Species Transport,” AnalyticalChemistry, 72, pp. 5473-5482(2000).
[20] Griffiths, S. K. and Nilson, R. H., “Low-Dispersion Turns and Junctions for Microchannel Systems,” Analytical Chemistry, 73, pp.272-278 (2001).
[21] Molho, J. M., Herr, A. E., Mosier, B. P., Santiago, J. G. and Kenny, T. W., ”Optimization of Turn Geometries for Microchip Electrophoresis,” Analytical Chemistry, 73, pp. 1350-1360(2001).
[22] CFD-ACE(U) User Manual, Version 6.4, CFD research corporation (2000).
[23] Duncan J. Shaw, Introduction to Colloid and Surface Chemistry, (1989)
[24] Russel, W. B., Saville, D. A. and Schowalter, W. R., “Colloidal Dispersion,” New York, (1989).
[25] CFD-ACE(U) User Manual, Version 5, CFD research corporation (1998).
[26] 黃福居, ”全三維軸流風扇的葉片最佳化設計,” 國立成功大學機械研究所碩士論文(2001).
[27] 葉怡成, ”類神經網路模式應用與實作,” 儒林圖書有限公司(1998).
[28] 呂卓穎, ”神經網路之探討與應用,” 國立成功大學化學工程研究所論士論文(1997).
[29] 劉惟信, ”機械最佳化設計,” 全華科技圖書公司(1996).
[30] 李生丕, ”應用類神經網路於軸流風扇葉片設計,” 國立成功大學機械研究所碩士論文(1999).