An Adaptive Disturbance Observer for Precision Control of Time-Varying Systems*

In this paper, a new adaptive disturbance observer (ADOB) approach is introduced to overcome the inherent interference problem between a disturbance observer (DOB) and a parameter adaptation algorithm (PAA); the PAA is applied to estimate the parameters of the nominal plant model inverse, such that the disturbance estimated by the DOB is remained small. Namely, the proposed ADOB is equivalent to an adaptive control method that recursively minimizes the ℓ2 norm of the disturbance estimated by a DOB. Since the bounded ℓ2 norm of the disturbance estimate can be interpreted as the guaranteed stability of the DOB loop, the stability of the proposed ADOB can be proved theoretically, as well as practically. In addition, the proposed method directly seeks the parameters of a nominal plant model inverse, and thus the model inversion process is no longer necessary. The proposed ADOB is verified by theoretical analyses and experimental results in this paper.