The conception of electromechanical oscillations initiates in the power network when there is an installation of the generator in parallel with the existent one. Further, the interconnection of multiple areas, extension in transmission, capricious load characteristics, etc. causes low-frequency oscillations in the consolidated power network. This paper proposes variants of a booming population-based grey wolf optimization (GWO) algorithm in the tuning of power system stabilizer parameters of a multi-machine system in damping low-frequency oscillations. The parameters have been tuned by framing an objective function considering the improving damping ratios for the system states with lesser damping ratios and shifting the system eigenvalues towards the left-hand side of s-plane for the improved settling characteristics for the oscillations in the system. The requisites of stabilizer strategy are mapped with the hallmarks of prevalent algorithms and designed hybrid versions of GWO for the enhancement of the multi-machine power system stability. Four variants of GWO technique are nominated based on the competent stabilizer performance namely, modified grey wolf optimization (MGWO), hybrid MGWO particle swarm optimization (MGWOPSO), hybrid MGWO sine cosine algorithm (MGWOSCA) and hybrid MGWO crow search algorithm (MGWOCSA) for the designed multi-machine power network. The proposed methods have been realized with the statistical analysis on the 23 benchmark functions. Nonparametric statistical tests, namely, Feidman test, Anova test and Quade tests, have been performed on the test system, further analysed in detail. A detailed comparative analysis under the self-clearing fault is presented to illustrate the suitability of the proposed techniques. For the analysis purpose, the location of system eigenvalues has been observed along with their oscillating frequencies and corresponding damping ratios. Further, the damping nature offered with considered system uncertainty for the system states also presented with the PSS parameters obtained by the proposed algorithms.
Read full abstract