むだ時間を考慮したニューラルネット制御による油圧サーボモータ系の研究

概要

論文の詳細を見る
In the previous papers, it was clarified that the application of the dither signal into a hydraulic servomotor position control system using a Neural Network is effective in reducing the steady error. On the other hand, the results of the transient characteristics were not satisfied, when the step input signal is provided. In this report, methodology, which improves by cooperation between the reference model and a delay element, is proposed. The suitable non-dimensional dead time can be analytically obtained from simulated response waves in a state feedback control system. As a result, it is experimentally verified that the position over-shoots and errors can be minimized if the dead time is appropriately chosen. Furthermore, the effectiveness is certainly confirmed even though the experimental parameters, such as supply pressure, load inertia, and setting angular position input are varied.
社団法人　日本フルードパワーシステム学会の論文
2006-01-15