Abstract
This paper studies a method for designing a robust servo system in which the plant variables (i.e., plant input and plant output/state) converge to their steady-state in an optimal fashion with respect to the usual regulator-type quadratic-integral performance index. Since the performance index does not weight the compensator state, the optimality attained by the resulting control system has a definite significance that it achieves quick tracking of the plant output/state while suppressing a large peak in the plant input. Also, the steady-state error compensation involved in the resulting control system has a desirable feature that it actually works only when there exist some disturbances and/or modeling error of the plant. Accordingly, this design method provides us with two-degree-of-freedom approach, in which we can tune feedback characteristics such as disturbance attenuation and robust stability without changing the responses of the plant input and plant output/state for the reference input at all.
