Nominal model selection for robust control design

The standard statement of equivalent models between state space and transfer matrix representation assumes the absence of uncertainty in the model. Here, the effects of the unequivalent representations under uncertainty are explored when an H/sub /spl infin// controller is designed. The robustness analysis is made taking, from a missile autopilot data, four different linear nominal models with uncertainty, which are obtaining in essence by the same characterization. It is shown that the nominal model must be considered as a tool to simplify the design task and its selection plays an important role to satisfy the performance for the family of plants.