ANN Based Method for Power System Harmonics Estimation

Estimation of power system harmonics has become a real challenge for power system researchers due to proliferation of power electronic based devices. This paper proposes a neural network based approach to detect and estimate the harmonics present in a signal. The harmonic information in a signal is extracted using fast Fourier Transform based power spectral density (PSD) calculation. The PSD values are fed to the neural network model which accurately estimates the amplitudes of harmonics present in the signal. The method is simple and can be used for estimating multiple harmonics present in a signal. The effectiveness of the proposed method has been tested on a rectifier circuit.