一般伺服機構若希望其轉速能追循任意之軌跡，以目前常用線性之控制器是不易達成的，而反覆學習控制理論能夠藉由重複疊代學習，將誤差量降低，以達成控制目標。但反覆學習控制在使用上有其限制，其中最容易造成學習失敗的因素便在於外界的振動、干擾等。因此，在此將輔以離散小波轉換(Discrete Wavelet Transform, DWT)以分離不可學習之成分，可學習之部分由學習控制處理，而不可學習之部分則以回授控制壓抑之，使反覆學習控制更具強健性。本文以基於小波轉換之反覆學習控制使伺服機構達成追循頻寬下任意軌跡之目標。
本文以具外界干擾與參數不確定之線性非時變系統為主題，設計一基於小波轉換之反覆學習控制器，提出學習控制器參數之設計法則、討論離散小波轉換在外界干擾下對反覆學習控制效能之影響、以及感測器解析度對學習控制收斂性之影響等。然後更建構硬體之實作平台以及其周邊電路，完成即時硬體迴路，並於數位信號處理器(Digital Signal Processor, DSP)上實現此控制律，將此控制律應用於真實之直流伺服馬達上，最後將實驗結果與模擬結果相互比對，以驗證基於小波轉換之反覆學習控制律比一般反覆學習控制更具強健性，並使伺服馬達能夠經由反覆學習追循其頻寬限制下之任意速度軌跡。

It is not easy for a typical servo mechanism to track arbitrary velocity profile by linear controllers. Iterative learning control (ILC) scheme can reduce the tracking error through repeated trials to achieve this goal. There are restrictions for using ILC, certain dynamics, e.g. vibrations and disturbances will cause the learning task to fail. Here, a discrete wavelet transform (DWT) is employed to extract unlearnable dynamics. While learnable dynamics is handled by ILC, unlearnable dynamics is regulated by feedback control. It will let ILC become more robust. Within the operational bandwidth, applying DWT based ILC can achieve excellent speed-tracking performance of the servo mechanism under study.

This thesis presents a DWT based Iterative learning controller for DC servo motor. An algorithm to design learning gains and feedback gains that guarantees the convergence of learning curve is proposed. Its feasibility is verified through simulations. The effect of disturbances and sensor resolution on the performance of ILC is explored. Experimental platform and the peripheral circuit were devised. ILC is implemented on a Digital Signal Processor (DSP), and the control law is realized on the DC servo system. The experimental results further verify the simulated results. According to the experimental results, DWT based ILC is more robust than general ILC and can let DC servo system to track arbitrary speed profile within the operational bandwidth by iterative learning.

[1]M. Vetteri, and C. Herley, ” Wavelets and filter banks: theory and design,” IEEE Transactions on Signal Processing, Vol.40, No. 9, pp. 2207 -2232, 1992.
[2]G. Strang and T. Nguyen, Wavelets and filter banks, Wellesley-Cambridge Press, 1996.
[3]單維彰,凌波初步,全華科技圖書公司,2000年
[4]S. Arimoto, S. Kawamura, and F. Miyazaki, ” Bettering operation of robots by learning,” J. of Robotic Systems, Vol. 1, pp. 123-140, 1984.
[5]T. J. Jang, C. H. Cho, and H. S. Ahn, “Iterative learning control in feedback system,” J. of Automatica, Vol. 31, No. 2, pp. 243-248, 1995.
[6]S. Arimoto, T. Naniwa, and H. Suzuki, “Robustness of P-type learning control with a forgetting factor for robotic motions,” Proc. 29th IEEE Conf. Decision and Control , Dec. 5-7, Vol. 5, pp. 2640-2645, 1990.
[7]L. Cai, and W. Huang, ”Fourier base learning control and application to position table,” Robotics and Autonomous System, Vol.32, pp. 89-100, 2000.
[8]D. Huang and Y. Jin, ”The application of wavelet neural networks to nonlinear predictive control,” IEEE International Conference on Neural Networks, Vol.2, pp. 724 -727, 1997.
[9]Kune-Shiang Tzeng, and Jian-Shiang Chen, " Toward the Iterative Learning Control for Belt-driven System using Wavelet Transformation", to appear in the Journal of Sound and Vibration, 2004.
[10]J. X. Xu, R. Yan, and Z. H. Guan, ”Direct learning control design for a class of linear time-varying switched systems,” IEEE Transactions on Circuits and Systems, Vol.50, NO. 8, pp. 1116 -1120, 2003.
[11]ADSP TS101S DSP Data Sheet, 2002.
[12] ADSP TS101S EZ-KIT LITE evaluation system manual, 2002.
[13]ADSP TS101S hardware manual, 2002.
[14]王信博,光碟機之主軸伺服控制晶片之設計與實作-國立清華大學動力機械工程學系碩士論文,2000年。