Robust control of multivariable systems using statistical confidence bounds

It is well known that traditional algorithms from the system identification field produce, not only a nominal model, but also a quantification of the quality of the model via statistical confidence bounds. However, traditional robust control design methods often do not match well this type of description of modelling errors. This paper extends a recently developed procedure for dealing with statistical confidence bounds in robust control to multivariable systems. The solutions are obtained in the spirit of model matching via a state-space approach. A design example is introduced to demonstrate the efficacy of the method