HHT Improve Power Quality Disturbance Signal Analysis Applications

In this paper, a novel method for automatic power quality disturbance classification is presented. It’s based on intrinsic modal feature energy and artificial neural network. Gain a set of the intrinsic mode function(IMF) and a residue by using empirical mode decomposition(EMD) to decompose the disturbance signal, during the procession of EMD. end effects are overcome hy extension and plus windows. Secondly. extract the feature vector from IMFs Then get the training and testing sample set which include time and frequency of disturbing, train the BP neural network and achieved disturbance classification. Simulation results show that the proposed method can class the non-stationary power disturbance signal accurately, such as voltage sags, swells, shor. interruptions, voltage spikes, transient harmonic and oscillation and so on.In order to overcome the shortcoming of EMD and EEMD. this paper presents a new method of refined Hilbert-Huang transforms (HHT) to analyzt power quality disturbance. Both end of disturbance signal are predicted hy artificial neural network. then plus windows to the signal which is extension, the end effect of EMD decomposition can be inhibition effective. Then disturbance signals are decomposition by the method of mask signal which is based or fas Fourier transform (FFT). a serious of intrinsic mode function (IMF) are obtained then amplitude and frequency of harmonic can be obtained, and get it derivative to determine the moments when the disturbance occurred and endedThe data collected from actual power quality disturbance signal are very large, especially in the case of high-order harmonic. traditional sampling methods require a high sampling frequency. which wili produce a large amour(?) of data, and it’s not conductive to the transmission and processing. Therefor compressed sensing theory is introduction to power quality disturbance signal processing. sampling and compression data at the same time. which can reauces the hardware cost of the sampling largely.