A parametric LQ approach to multiobjective control system design

The synthesis of a constant-parameter-output feedback control law for a constrained structure is set in a multiple-objective linear quadratic regulator framework. The use of intuitive objective functions such as model-following ability and closed-loop trajectory sensitivity allows multiple-objective decision-making techniques, such as the surrogate-worth tradeoff method, to be applied. For the continuous-time deterministic problem with an infinite time horizon, dynamic compensators as well as static output feedback controllers can be synthesized using a descent Anderson-Moore algorithm that is modified to impose linear equality constraints on the feedback gains by moving in feasible directions. Results of three different examples are presented, including a unique reformulation of the sensitivity reduction problem