Analysis of harmonic detection algorithms and their application to active power filters for harmonics compensation and resonance damping

In this paper, the performance of four harmonic detection methods is evaluated in terms of accuracy, speed of convergence, computational complexity and memory requirements; operation with measurement noise and variations of the signal amplitude and fundamental frequency is considered. The harmonic detection algorithms are based on the discrete Fourier transform (DFT), the recursive discrete Fourier transform (RDFT), the Kalman filtering (KF) approach, and the instantaneous reactive power (IRP) theory. Results obtained by simulation with MATLAB/Simulink and their real-time validation with the dSPACE simulator are presented to compare the detection methods. The effectiveness of the algorithms is demonstrated in their application to the control of an active filter and a hybrid active power filter dedicated respectively to harmonics compensation and to harmonic resonance damping in industrial power systems.