Design of l1-optimal controllers by multiobjective linear programming

The design of discrete-time controllers as a multiobjective l₁-optimal control problem is addressed in this paper. The multiobjective linear programming problem that results from the use of approximation methods based on the theory of l₁ optimization is solved by a goal programming technique, which reveals to be an efficient design technique. Performance specifications are easily described as goals for the objectives and their trade-offs can be explicitly represented. In addition, the goal programming formulation efficiently explores the properties of l₁-optimal control methods based on linear programming. The paper includes numerical examples that illustrate the proposed approach