Study On Harmonic Detection & Suppression In Power System Based On Anti-aliasing Wavelet Theories

Power system harmonic issue is a challenge topic in the world. As core of harmonic issues, harmonic detection and suppression is the main basis and final goal of harmonic studies. Owing to the inherent nonlinearity, randomness, distributing, nonstationarity and complexity of influencing factor on power system harmonics, many conventional detection techniques can't acquire accurate harmonic identification, so it is excellent valuable in theory and practice to investigate new detection and suppression methods. Because of excellent time-frequency localization characteristics, wavelet transform (WT) can be used to detect harmonics. Since IEEE proposed applying WT to harmonic analysis in the 7^th international harmonic conference in 1996, WT based methods gained rapid development in harmonic detection. However, owing to aliasing within almost all the existed wavelets, even though it is shown that wavelet-based methods are superior to the traditional detection techniques, they can't satisfy the accurateness and robustness requirement. This dissertation deeply studies harmonic detection and suppression technology based on anti-aliasing wavelet theories. Author's main work concentrates on three aspects as follows:(1) Wavelet aliasing error estimation and aliasing reason analysisWT total error and aliasing error is estimated by using Shannon sampling theory and Walter sampling theory, and aliasing cause is clarified in many ways such as physics essence, magnitude-frequency characteristics and filter design limitation. In addition, aliasing phenomenon and aliasing harm is illustrated.(2) Three anti-aliasing wavelet based harmonic detection methodsâ‘  Harmonic detection method based on frequency-domain interpolation anti-aliasing wavelet transform. Shannon wavelet function which can divide frequency-band strictly is selected as generating function; the frequency-domain interpolation algorithm is used to cancel wavelet aliasing and Gibbs phenomenon; Chirped Z transform is used to achieve fast computation.