Control-oriented approximate closed-loop identification via fractional representations

We propose an iterative solution to the combined problem of approximate identification and control design. The characteristic behaviour of a plant operating under stabilizing feedback is approximately modeled. The resulting model is used to carefuily redesign the compensator such that a better performance is achieved without loosing the validity of the model. The new compensator is applied and subsequently the above sequence is repeated. A case study shows that the compensator corresponding to the eventual model controls the plant better than ever would be possible by the best approximate open-loop model.