Comparison of real-time harmonic estimation algorithms

Abstract

The increasing use of nonlinear loads and power-electronic converters has led to growing harmonic distortion in modern power systems, creating a need for accurate and fast harmonic estimation methods. This thesis compares three techniques: the Fast Fourier Transform (FFT), the Multiple Second-Order Generalized Integrator (MSOGI), and the steady-state Kalman filter (SSKF). The methods are evaluated using MATLAB simulations in terms of settling time, accuracy, and computational load. The results show that the SSKF achieves the fastest dynamic response (≈ 10 ms), followed by the FFT (20 ms) and MSOGI (≈ 60 ms). The MSOGI has the lowest computational load when fewer than five harmonics are tracked, while the FFT becomes more efficient when a larger number of harmonics is considered. The SSKF is significantly more computationally demanding in all scenarios. These results highlight trade-offs between speed and computational efficiency in real-time harmonic estimation.