Estimation of harmonic currents injected by nonlinear loads for a distorted power supply scenario using Artificial Neural Networks

The determination of the true current harmonic distortion produced by a nonlinear load is problematic due to the fact that the voltage waveform provided by the power supply at the Point of Common Coupling (PCC) rarely is free of harmonic distortion. This paper proposes a method based on Artificial Neural Networks (ANN) in order to realize the estimation of the harmonic content presented in the nonlinear load current waveform supposing a pure sinusoidal power supply. However, this estimation is made when the nonlinear load is in fact connected to a distorted power supply. For this, a dynamic ANN with time delays is utilized for modeling the nonlinear load admittance in a pre-established condition of PCC voltage with little or none harmonic distortion. Once achieved a satisfactory training, now for a PCC voltage with any level of harmonic distortion, the ANN is utilized for predicting the nonlinear load current waveform by a power supply estimation that preserves similarities with that applied during its training. One of the advantages of the method is that only the instantaneous values of voltage and current are necessary for modeling the loads. It does not need the knowledge of any load or power system parameter.