抄録
This paper proposes an interesting technique for LMI-based multiobjective control design. We deal with strictly positive real H2 controller synthesis. If we partially fix both LMI solutions and controllers in the first step, a set of jointly convex LMI formula is obtained in the second step. The first step restricts the range of the parameter set to some degree, and the second step searches within these boundaries. However, if these constraints include the standard result for a common LMI solution, then the second step may improve this result while allowing uncommon LMI solutions. A numerical example is included.