Input Output Selection Based on Worst Case Uncertainty

Selection of control actuators and sensors aimed at Robust Performance (RP) is an important issue. This Input Output (IO) selection could be based on p-synthesis, which is generally time-consuming. The basic idea of a new and more efficient IO selection method is to get rid of the structure of the combined uncertainty/performance block A arising in RP control problems. For this purpose, the concept of Worst Case Uncertainty (WCU) is introduced. This is the smallest uncertainty (in Xm norm sense), that violates the considered RP requirement. The IO selection method proceeds as follows. First, an optimal p-synthesis is performed including all actuators and sensors (full IO set). Second, for the closed-loop, WCU data is generated across a frequency grid. Third, a representative, real-rational, proper, and sta-ble transfer function matrix is constructed for this data. Fourth, this (possibly structured) WCU representation is absorbed into the generalized plants for other IO sets and the struc-tured A reduces to an unstructured performance block. Finally, IO selection amounts to checking criteria for the existence of a stabilizing controller achieving a specified closed-loop Xm norm. From previous research it is known, that this can be performed efficiently