Direct Variance Design - A Multiobjective Control Theory

This paper offers a theory for designing linear state feedback controllers to directly accommodate variance constraints imposed on the system states. Such controllers guarantee an "average" closed-loop response to gaussian input disturbances. The method provides a mechanism for directly assigning all variance values (the diagonal elements of the covariance matrix) to the closed-loop system. The theory may be viewed as offering direct access to a new set of design variables; the state variances.