Controller design of decentralized singularly perturbed systems

Abstract

In this thesis, the performance of a decentralized system was analyzed based upon the design of the controller, either a reduced-order model or a full-order model. The singular perturbation technique was used to obtain the reduced-order model of a decentralized system. Then this model was used to design a controller for state feedback. Since this controller was the reduced-order model of the system, it was implemented based on the full-order model by padding zeroes. Thus, the controller was an approximation. Hence, a performance analysis was conducted to verify the near approximation of the design using the singular perturbation technique. To check the sensitivity of the design, the performance of the system using the controller designed on the full-order system was compared with the performance of the system using the controller designed on the reduced-order model. A comparison of the results showed that the effectiveness of the reduced-order model design can be verified and checked as to whether it is giving the required approximate results for a lower-cost controller design.