Experimental demonstration of disturbance suppression control with novel nonlinear disturbance predictor based on reconstructed attractor

In this paper, disturbance suppression control with a novel nonlinear disturbance predictor to be used for motion control system is proposed. By design of the novel disturbance predictor, we develop disturbance observer to realize disturbance feedback without time delay. This predictor is built from the reconstructed attractor which is often used in chaos analysis. The reconstructed attractor consists of the time series of disturbance and expresses the dynamics of disturbance. Due to the characteristics of the reconstructed attractor, it is not necessary to identify or model the dynamics of disturbance. Therefore the proposed method is especially effective to compensate the interference force of multiple axes robot manipulator friction force and so on which control tends to be complicated in the conventional methods. In the proposed method, one-step-ahead prediction can be realized by extension of the trajectory in the state space of disturbance intuitively. But in the reconstructed attractor, the order of the space can be selected arbitrarily as prediction is succeeded. Additionally the state transition equation of disturbance is not necessary. To verify the effectiveness of the proposed method, some simulation results using a double pendulum system are shown. In this simulation, though the chaotic interference force gives bad influence to the system, the proposed method can be succeeded in stabilizing the behavior of the double pendulum.