Power System Stabilizer based on a Neural Supervisor from a Local Model Network

C. Tavares-Da-Costa, Jr., J.A.L. Barreiros, A.M.D. Ferreira, and W. Barra, Jr. (Brazil)


Power System Control, Power System Stabilizers, Neural Networks, Local Model Networks, Excitation Control.


This work proposes the design of a Power System Stabilizer using a Neural Supervisor trained from a Local Model Network. For any given system operation point, represented by the active and reactive power in the synchronous generator output, the neural supervisor can supply a set of parameters that completely define a linear model of the power system. Then, a pole-placement method is used to obtain a discrete-time controller to be used in the actual system operating condition, improving the power system dynamic stability. Simulation results confirm the good performance of this approach when compared with a fixed-parameter PSS and a self-tuning stabilizer.

Important Links:

Go Back