Use of intelligent methods in harmonic analysis of power systems

In the last decades, with the increasing use of non-linear electronic equipmentsin power systems, harmonic pollution has become an important issue in powerquality. To prevent problems due to the harmonics and to improve the qualityof the delivered energy, estimation of harmonic parameters magnitude andphase angle, is an important task. Conventionally Fourier Transform basedalgorithms are the most commonly used techniques for harmonic estimation,however these techniques have certain limitations and drawbacks. On theother hand, recently several alternative algorithms utilizing intelligent methodshave been proposed for harmonic estimation. In this thesis, three hybrid algorithms,composed by combining Least Squares Method with evolutionary computationalgorithms, are applied in harmonic estimation. These algorithms, by utilizingthe feature that harmonic estimation problem is linear in amplitude andnonlinear in phase, use evolutionary computation algorithms for phaseangle estimation and Least Squares Method for amplitude estimation.Two of the hybrid algorithms using Genetic Algorithm and Particle SwarmOptimization for phase angle estimation, were introduced in literaturepreviously. In this thesis a novel hybrid algorithm which uses DifferentialEvolution for phase angle estimation, instead of Genetic Algorithm orParticle Swarm Optimization, is presented. The applications are realized insimulation environment and the results of the algorithms are compared.