Plant Controller Design by Convex Optimization

Dr. Carsten W. Scherer

In addition to the task of designing a dynamic controller in a typical control application, it often happens that certain plant parameters can be adjusted to favorably influence the behavior of the controlled system. The problem of specifying both a dynamic controller and real-valued parameters of the plant in order to optimize a chosen performance criterion is called the plant controller design problem. The topic of this talk is a novel plant controller design technique based on convex optimization if the controller and parameters enter the closed loop in a specific fashion. Motivating examples with clarify the range of applicability of these algorithms. As a specific model problem that is interesting in its own right we discuss the design of controllers that are robust against uncertain stochastic disturbances, what considerably generalizes both H2 and H-infinity control. We reveal how infinite dimensional Lagrange duality can be exploited to transform the corresponding robust analysis problem into a parametric optimization problem. This renders our algorithms amenableto design of robust output feedback controller without involving any conservatism.

Biographical Sketch:

Carsten Scherer received the diploma degree and Ph.D. degree in mathematics from the University of Würzburg(Germany)in 1987, Dr. Scherer spent six months as a visiting scientist at the Mathematics Institute of the University of Gronongen(the Netherlands). In 1992, he was awarded a grant from the Deutsche Forschungsgemeinschaft(DFG)for six months of post doctoral research at the University of Michigan(Ann Arbor)and at Washington University(St. Louis)respectively. In 1993 he joined the Mechanical Engineering Systems and Control Group at Delft University of Technology(The Netherlands)where he currently holds a position as an associate professor(universitair hoofddocent).His main research interests cover various topics in applying optimization techniques for developing new advanced controller design algorithms and their application to mechatronics and aerospace systems. Currently, Dr. Scherer is spending a three months sabbatical at the Automatic Control Laboratory at ETH Zürich.