An Experience Replay Based Adaptive Disturbance Observer for a Class of Nonlinear Systems

Abstract

This paper designs an adaptive disturbance observer for a class of nonlinear systems with unknown disturbances where the disturbance is assumed to be generated by some unknown dynamics. we first use a filtered regressor approach to model the nonlinear systems and the disturbance. We show that this filtered regressor form allows us to estimate the disturbance using only measured state. Next, to improve convergence speed we propose a new adaptive observer with experience replay to ensure that disturbance estimate error is globally exponentially stable. Experience replay uses past measured data to not only assure that the observer estimation error converges to zero with a rate depends on the Minimum eigenvalue of the history stack matrix. Compared to the existing results, we neither use the knowledge of the disturbance dynamics nor the state derivatives in our adaptive protocol. Finally, we use a simulation example to illustrate the effectiveness of our results.