A hybrid neuro-fuzzy power system stabilizer

Abstract

The paper presents a novel artificial intelligent (AI) neuro-fuzzy hybrid power system stabilizer (PSS) design for damping electromechanical modes of oscillations and enhancing power system synchronous stability. The hybrid PSS comprises a front end conventional analog PSS design, an artificial neural network (ANN) based stabilizer, and a fuzzy logic post-processor gain scheduler. The stabilizing action is controlled by the post-processor gain scheduler based on an optimized fuzzy logic excursion based criteria (J/sub 0/). The two PSS stabilizers, conventional and neural network have their damping action scaled on-line by the magnitude of J/sub 0/ and its rate of change (dJ/sub 0/). The ANN feedforward two layer based PSS design is the curve fitted nonlinear mapping between the damping vector signals and the desired optimized PSS output and is trained using the bench-mark analog PSS conventional design. The fuzzy logic gain scheduling post-processor ensures adequate damping for large excursion, fault condition, and load rejections. The parallel operation of a conventional PSS and a neural network one provides the optimal sharing of the damping action under small as well as large scale generation-load mismatch or variations in external network topology due to fault or switching conditions. >