Quantifying the Model Accuracy Needed for Control *

The issue of how to usefully quantify the amount of modeling accuracy needed to meet a given set of control specifications is discussed. The concept of H _∞ multiplicative error is shown to play a central role. References are listed describing recent theoretical advances in multiplicative-error model order reduction and system identification, and their application in design studies in which the methods were applied to challenging multivariable control problems involving a flexible space structure and a highly maneuverable fighter aircraft.