誤差訊號處理於學習控制性能之研究

Error Signal Processing in Iterative Learning Control

羅印隆(Lo Yin Lung)

指導教授 ： 陳世樂

國立中正大學/工學院/機械工程學系/碩士(2012年)

摘要

本研究是在命令式疊代學習下分析各種不同訊號處理法則用於誤差訊號所造成的效果，某些複雜的訊號處理法則可能使效果較好、簡單的訊號處理法則可能效果沒這麼好，但是也達到要求，所以在本研究也將這些結果統計出來，在往後根據不同的需求而選擇對應到的誤差處理法則於疊代學習控制中。而所謂的命令式疊代學習是應用在有週期性，反覆加工的控制法則，它是根據前一次加工中的誤差經由學習演算法則處理後再改變下一次的命令，這樣反覆學習使誤差越來越小。但因為不同的訊號處理會使疊代學習的效果不相同，所以在本論文將一些常用的誤差訊號處理分成兩部份，第一部份是疊代學習控制之性能提升方法(消除非重複誤差、局部強化)，第二部份是濾波器(小波轉換、一階低通濾波器、moving average、模態分解)，並將這兩部份應用於攻牙學習，在不同的進給速度下整理出在什麼條件的需求可使誤差收斂效果最好。實驗結果證實，經過7次以內之學習，主軸轉速於1000rpm可將同步誤差降低17倍、6000rpm可將同步誤差降低58倍，在高速進給下差異相當明顯。
本研究主要貢獻在於
1.將適合的濾波器與疊代學習控制之性能提升方法作整合使疊代學習的精度提高。
2.整理出針對不同的精度、速度要求，選擇較適當的誤差訊號處理方式。

關鍵字

疊代學習控制 ； 誤差訊號

並列摘要

This study analyzes effects of error signal using different signal processes.
Some complex signal processes may have better effects and some simple ones have effects which are not so good but achieve specifications,too.
This study shows the statistics results, so that we can choose the best error signal process in iterative learning controling.
Command-based iterative learning is a kind of controling law used in periodic and iterative tapping. It changes the next command by dealing with the previous error with the learing algorithm, and we can make error smaller and smaller in this iterative learning way.
But different signal processes will make different learning effects, so this study divides common error signal processes into two parts. The first one is performance improvement for iterative learning control and the other one is filters(wavelet transform,first order low pass filter,moving average,empirical mode decomposition) .We apply these two parts in iterative learning controling for a rigid tapping system and summarize what demand makes the best effect of error convergence in different speed feeding situations. The experiment results show that, we can reduce synchronous errors by 17 times with the spindle speed of 1000 rpm , 58 times with the spndle speed of 6000 rpm. The effect is pretty significant with high-speed feeding.
The main contributions of this study are:
1. We integrate suitable filters and performance improvement for iterative learning control to make iterative learning control results more accurate.
2. We summarize the way to choose the proper error signal process in different demands of accuracy and speed.

並列關鍵字

Iterative Learning Control ； error signal

參考文獻

[1]. Arimoto, S.,Kaeamura,S. and Miyazaki,F., 1984, ”Bettering Operation of Robots by Learning,” Journal of Robotic System,Vol.1(2), pp.123-140.

[2]. Moore, K.L., 1993, Iterative Learning Control for Deterministic Systems, London: Springer-Verlag.

[3]. Richard W. Longman, “Iterative learning control and repetitive control for engineering practice,” International Journal of Control,73:10,930-954

[5]. J. Hatonen, D.H. Owens, K. Feng, 2006,“Basis functions and parameter optimisation in high-order iterative ”, Automatica, Volume 42, Issue 2, Pages 287-294

[6]. W.B.J. Hakvoort, R.G.K.M. Aarts, J.van Dijk, J.B. Jonker, 2009,“A computationally efficient algorithm of iterative learning control for discrete-time linear time-varying systems ”, Automatica , Volume 45, Issue 12, Pages 2925-2929