Novel Harmonic Identification Algorithm based on Fourier Correlation and Moving Average Filtering

With the aim to minimize the harmonic disturbances created by non linear loads, the active power filters have an important role in the improvement of power quality. Active filters need to identify the harmonic components of some specific system currents/voltages to calculate the filter current/voltage references and thus set the power devices switching to compensate for the harmonics. Therefore, the accuracy and quickness in the harmonic identification of the input signals have a very important role in the active filters control performance. This work presents a novel digital algorithm for harmonic identification based on Fourier correlation in quadrature and moving average filtering (MVA). The input signal has harmonic component at well known odd frequencies, and it can be expressed as a Fourier series. Therefore, by multiplying independently two orthogonal unitary signals rotating at a known frequency by the input signal, the output signals have the Fourier coefficients of this frequency shifted to dc and the other coefficients shifted to even harmonics. Then by placing a low pass filter after each multiplier, the dc coefficients are obtained. This correlation method has the problem of a relatively high gain of the employed low pass filters for the lowest order even harmonics.