| New methods for decomposition and decentralized output control of large-scale systems are presented. An existing method for decomposing a complex system into weakly coupled subsystems is a graph theoretic technique called ;Model Reference Robust Control (MRRC) is an effective feedback control method for output tracking of uncertain single-input, single-output minimum phase linear systems. The thesis develops decentralized MRRC for use in output tracking of interconnected uncertain plants in the presence of disturbances. In addition, it is demonstrated that implementation of the control for plants of high relative degree is greatly simplified by placing a linear filter in parallel with the plant. The technique is applicable to both isolated plants and interconnected systems, and does not significantly affect tracking accuracy.;Decentralized MRRC used as output excitation control on synchronous machines is shown to be an effective means of damping oscillations in electric power systems. Unlike many other excitation control stabilizer designs, no fixed transfer function descriptions of the power system are required and coordination of decentralized controllers is unncessary. Simulations using a nonlinear model of a 39 bus, 10 machine system demonstrate the effectiveness of decentralized MRRC for stabilizing a power system in response to both small and large disturbances. |
