Robust power system stabilizer design.

Abstract

This thesis investigates the design of damping controllers to alleviate the problem of low frequency electro-mechanical oscillations in power systems. The operating point and network parameters of power systems are continually changing, resulting in changes in system dynamics. The conventional controller design methodology has therefore come under increasing scrutiny for its lack of considerations for robustness. The thesis first outlines the conventional design of a power system stabilizer (PSS) and then applies two robust techniques (Hoo and Quantitative Feedback Theory, QFT) to the design problem. The single machine infinite bus (SMIB) model is used to illustrate the procedure for all three design techniques. The final design is undertaken to illustrate the more important problem of robust multi-machine PSS design using QFT. The design requires linearised models of the multi-machine system. A brief discussion is given on how these can be obtained. An introduction to decentralized control design in QFT is included to support the multi-machine design. Chapter three proceeds through the design steps required to generate a conventional PSS. The technique is shown to be simple for a given set of operating conditions. The controller is shown to be adequately robust over the given set of operating conditions albeit not by design. Chapter four introduces a design technique that directly addresses robustness issues during the controller design. For a restricted range of operating conditions the designed controller demonstrates the desired robustness and performance characteristics. The inherent difficulties with Hoo in PSS design become more apparent as the operating range is extended. Chapter five introduces the second robust controller design technique. QFT is shown to be more adept at dealing with increased operating ranges and changing specifications in the single-machine infinite-bus case. The controller is easy to generate and performs well over the entire range of operating conditions. QFT is also applied to the controller design for a four-machine study system. The design is a marginally more complex than in the single machine case but is still easily accomplished. This thesis confirms previous attempts at solving the design problem using the methods outlined above. The performance of all controllers is assessed for small and large disturbances using non-linear time domain simulations with models developed using PSCAD/EMTDC and MATLAB.