HHT Algorithm And Its Application In Power Systems

Hilbert-Huang transform is a new time-frequency analysis method. It is appropriate in analysis of nonlinear and nonstationary signals, and has lots of advantages which adaptivity and unlimited by Heisenberg uncertainty principle. However, this method has shortcomings and deficiencies. The paper analyzes and researches the problems including end effect, curve fitting, termination criteria, illusive components and mode mixing. And the improved HHT is applied to parameter extraction of harmonics, voltage flickers and low frequency oscillations in power systems.Firstly, analysis methods of power harmonics, voltage flickers and low-frequency oscillations in power systems are introduced, and the advantages and disadvantages of existing methods are described. The existing time-frequency methods analysis and research a mutation signal, which verifies that Hilbert-Huang transform is better in analysis of non-stationary signals.For the end effects of HHT, this paper analyzes the direct extension method, mirror extension method and neural networks extension method, and proposes a new extension method which is called RBF-point symmetric extension method. It extends the signal to near zero by RBF neural network, and uses the point symmetric extension. Simulations show that the method can restrain end effects.The paper researches the problems including curve fitting, termination criteria, illusive components and mode mixing. To reduce the impact of sampling frequency, the sampling frequency is 8 times greater than maximum frequency and the whole periodical sampling. To avoid overshoot and undershoot phenomenon of cubic spline interpolation, hermite interpolation is used for the curve fitting, and the interpolation can inhibit mode mixing. In order to eliminate illusive components, the frequency number obtained by FFT is proposed to be decomposition termination conditions. This method can identify and remove illusive components, and reduce the computation because it is achieved before the appearance of illusive components. Improved criteria proposed by Rilling and the frequency numbers obtained by FFT are used to be termination condition.In order to overcome shortcomings and lack of the existing analytical methods, applications of HHT in the power harmonics, voltage flickers and low frequency oscillations in power systems are focused in this paper. Harmonic signal is preprocessed by an adaptive notch filter, and analyzed by HHT. This approach can effectively suppress the mode mixing resulting from too large fundamental frequency amplitude, and get high accuracy. Synchronous detection methods and low pass Butterworth filter are used to process voltage flicker, and HHT is used to analyze amplitude-modulated wave. Simulation indicates that the method can accurately extract frequency and amplitude from the single frequency and multi-frequency voltage flicker. Using time-varying and stationary low frequency oscillations in power systems signal to test and verify HHT algorithm, results of which show that it is very suitable for extraction of low frequency oscillations.