Comprehensive design methodology of tuned passive filters based on a probabilistic approach

摘要

This study presents a new method for the optimal design of single-tuned harmonic passive filters considering the probabilistic nature of power system magnitudes. For this purpose, the algorithm uses the measurements acquired by a power quality analyser. The Gaussian mixture model has been applied to actual harmonic measurements to obtain a precise statistical representation which is crucial in the proposed design procedure. The optimisation algorithm considers both probabilistic and deterministic variables involving all the harmonic components including the minor or non-characteristic harmonics. Moreover, the objective function of the optimisation algorithm is formulated taken into account all the harmonic terms overpassing other simplified formulations based only on the fundamental component. This study includes a case study with different optimisation scenarios to clearly highlight the benefits of each of the proposed contributions.