撓性旋轉連桿之類神經網路控制

Neural Network Control of a Flexible Slewing Link

郭中豐 ； 袁國獻

《中華民國振動與噪音工程學會論文集》 1996 (1996/04) Pp. 299-318

https://doi.org/10.30028/CSSVANNUAL.199604.0299

摘要

本文係運用類神經網路控制器以倒傳遞(Back Propagation)演算法爲學習法則，作撓性旋轉連桿之軌跡控制。首先利用Hamilton原理導出系統非線性動態方程式，以根軌跡法分析及設計出適當的加權值修正參數，結合假設模態法(Assumed-mode method)導出對時間相依函數之二階常微分方程式，並使用Runge-Kutta數值分析法，模擬所設計之類神經網路控制器作用下的系統及干擾暫態響應。
結果顯示運用類神經網路控制器具有自動調適外界干擾及減少系統振盪，並保有系統絕對穩定性之優點，確實達到撓性旋轉連桿反應速度快，準確度高之要求。

並列摘要

Neural-network Controller with the learning rate of Back Propagation is presented to control Flexible Slewing Link Robot. The first part of the project is to derive the equations of the flexib slewing link from Hamilton principle. The extended root-locus is used to analyze for Neural-networ exact weights. Next, in order to find the result of the distributed parameter system, the Assume mode method is adopted and an Neural-network Controller is formulated using state-spa representation of the dynamics of the system.
It can be seen that the designed Neural-network Controller not only can reduce the steady-staerrors, but also reduce the effects of disturbance of the system.

並列關鍵字

Flexible Slewing Link Robot ； Neural-network Controller ； Back Propagation