Transient Power Quality Detecting Based On Hilbert-Huang Transformation

As a kind of just emerging method of time-frequency analysis, Hilbert-Huangtransform can analyze non-linear and non-stationary signal. The method has highresolution, and the intrinsic mode function (IMF) components decomposed byempirical mode decomposition (EMD), has certain physical significance. Because ofthe advantagement mentioned above the method is very applicable to analyze non-stationary signal whoes frequency is variable with time.First of all, the basic theory of Hilbert-Huang transform is introduced in thispaper. Based on the expression of frequency and definition of instantaneous frequency,intrinsic mode function is discussed which makes instantaneous frequence hasvaluable significance, HHT’s completeness and local orthogonality are analyzed andfinally, main problem existing in HHT is pointed out.Furthermore, the reason, in this paper, of end effects existing in HHT method isanalyzed, based on which support vector machine prolongation method is proposed.Simulations are made to compare with five conventional methods, which proved thepossibilities of support vector machine prolongation method. Last, this method is usedto analyze power system’s multi-band harmonic. Experiment results demonstrate thatthe method can get clear instantaneous frequency accurately and reflect signal’s time-frequency characteristics, effectively repress effects and improve HHT’s effects.In addition, after summarizing the existing filtering method and in the hoping ofsolving the problem of denoising existing in practical engineering, this paper proposesan improve method-HHT spectrum filtering based on wavelet threshold processing,which not only can directly filter noise from HHT spectrum, but also can retainsignal’s nonlinear and non-stationary characteristic as much as possible. Comparedwith EMD filter, wavelet threshold filter, wavelet threshold filter based on EMD,Hilbert spectrum filter based on EMD as well as their signal-to noise both before andafter the filter, the method proposed in this paper get the best results. At last, the proposed method is used to analyze actual power signal, which proves this method willhave wide application prospects in actual engineering signal.