本計畫主要針對馬達驅動之運動系統，使用干擾觀測器(Disturbance Observer)架構，該控制架構主要使用驅動器之扭矩迴路，再自行架設速度迴路之PI控制器，由馬達輸出之轉速作為DOB之輸入訊號，而DOB之架構為參考模型第一階的倒數，DOB所得之輸出扣除輸入之訊號再經由一個有增益值的積分器回授於輸入端補償，使馬達轉動達到預期之結果，其中選用Butter worth低通濾波器，濾除輸出轉速之不必要雜訊使得干擾觀測器在實驗平台上得到有效的結果。該控制方案可以減少由參考模型和未知的系統模型之間的差異所造成的不確定性與摩擦力。 另外提出一個消除系統由非線性因素所造成之自我激發振盪(hunting)之抑制控制架構與控制命令平滑化架構，自我激發振盪控制架構是用於速度迴路，並使用類神經演算法調整出該控制抑制架構Ks 、Ka值，使得振動抑制控制架構可以有效抑制此系統所造成的振盪，控制命令平滑化架構用於位置迴路，增加前饋參數使馬達輸出結果達到優化，再使用基因演算法調整前饋參數得到最佳化參數。 ;In this paper, the motion system with a motor drive using a disturbance observer is considered. The control architecture is mainly applied to the torque loop of the drive circuit. The signal of the motor speed is fed to the disturbance observer as an input signal. The disturbance observer consists of the inverse of a reference model. In addition, a Butterworth low-pass filter is used to filter out noise, so the disturbance observer can achieve effectively results in the experimental platform. This control scheme can reduce the effect due to friction and modeling uncertainty. A hunting suppression framework to eliminate self-excited oscillation (hunting) of the control system is also proposed. A genetic algorithm is applied to adjust the values of control gains, so that optimal performance can be achieved.
