강인제어 기반의 비선형 외란관측기를 이용한 2축 매니퓰레이터의 궤적추종성능 개선

Abstract

This study have been performed in order that the active protection system that is mounted on the tank precisely aim the Anti-Tank Threats. Active Protection System is the form of a 2-axis manipulator, and is a non-linear systems that require very fast response speed and high precision. Therefore, in consideration of these characteristics, we used Sliding Mode Control and Nonlinear Disturbance Observer in order to show the exact tracking performance in both tracking the target trajectory which is fixed and tracking the target with time-varing. The result of applying the conventional disturbance estimation algorithm in target trajectory of time-varing as Sine waveform have showed the difficulty of the disturbance estimation and exacerbated rather the results of control. Because conventional disturbance estimation algorithm is based on a model of dynamic properties and the displacement, velocity, acceleration, we have found the cause of malfunction in the cumulative error of the differential term by frequency characteristics of the trajectory. To offset this, we devised a gain function that reflects the frequency characteristics and proposed disturbance estimation formula of the modified form. The proposed disturbance estimation equation properly estimated even disturbance in the target trajectory of time-varing, it can be confirmed that the steadily track the target trajectory and robustness by disturbances. Limit point of this study is that it was able to immediately apply the premise frequency gain function along the trajectory with knowing the frequency characteristics of the target trajectory in advance. It is needed that study about reflecting the frequency characteristics of the target trajectory in estimated disturbance even in environment that can not be known the frequency characteristics of the target trajectory. Further, it is necessary to study whether disturbance estimation is done properly in harsh environments where the frequency characteristic can not be figured.