Reduction of the Error Norm in Structurally Balanced Truncation

Abstract

In this paper, we attempt reduction of the error norm in structurally balanced truncation(SBT). SBT is carried out by solving two linear matrix inequalities(LMIs). However, it is difficult to obtain two optimal LMI solutions because the optimization problem has the objective function which is not convex. Therefore, we solve the optimization problem by approximating this objective function with a linear function, and we verify the validity of the proposed method.