Design of a Reduced-Order Disturbance Observer with Model Uncertainty

In this paper, we study the design of a reduced-order disturbance observer (DOB) controller for robust stabilization of systems with model uncertainties. In previous research, the high-frequency gain of the nominal model had to be carefully selected so that system stability is guaranteed. However, in this study, we propose a new Q-filter and show that, by employing it, the system can be stabilized even when an arbitrary high-frequency gain is used. The autonomy of such a design can be effectively utilized to enhance control performance, such as disturbance attenuation. From the simulation study, we verified an enhancement in disturbance-rejection performance: MAPE decreased from 4.44% (conventional controller) to 1.68% (proposed controller).