Gradient based hybrid metaheuristics for robust tuning of power system stabilizers

Abstract

Power System Stabilizers are controllers installed on synchronous generators to damp power system oscillations through the excitation control. These controllers can have either a conventional fixed structure composed by stages of gain and phase compensation or a flexible modern structure composed by three bands that correspond to a specific frequency range (low, intermediate and high frequency) in which each band is composed by two branches that are based on differential filters (with a gain, lead-lag blocks and a hybrid block). Power system stabilizers design is a hard and time consuming task and an alternative for tuning controllers is by using optimization methods. This paper presents three hybrid metaheuristics for the robust and coordinated design of power system stabilizers. The tuning procedure is modeled as an optimization problem which aims at maximizing the damping ratio coefficients in closed-loop operation. Robustness requirement is met by using multiple operating scenarios in the design stage. For solving the optimization problem, three metaheuristics (Gravitational Search Algorithm, Bat Algorithm and Particle Swarm Optimization) are combined with the Steepest Descent Method for local search capability enhancement. The proposed hybrid algorithms are applied to benchmark systems for validation.