Research On Power Quality Monitoring Signal Analysis Algorithms

Rapid technological and social development and improving living standards for electric utilities a higher power supply requirements, Modern power requirements are safe, effective, sustainable power supply, especially in a large number of high-precision, non-linear loads in power system applications on the quality of a higher power requirements.This article is presented for electric power quality signal analysis method based on Hilbert-Huang Transform harmonic "pollution" problem. Hilbert-Huang Transform has better nonlinear processing capability, and can handle a variety of complex power quality disturbance signals.Aiming at the complex diversity of power quality signals using highly nonlinear signal processing capabilities of Hilbert-Huang Transform(HHT) transformation, First point of power quality analog signal EMD decomposed IMF components with different frequencies and a residual component. Hilbert transform(HT) to the IMF components for each band, the time domain can be obtained by calculating the voltage magnitude diagram and time-frequency diagram. The simulation also get the starting and ending time of the disturbance Figure happened and disturbance information, and through experiments verify the accuracy of the information.Automatic recognition of power quality in recent years have proposed a hotspot, collecting data through predictive analytics can effectively determine the stability of the system. In this paper, automatic identification based on HHT and relevance vector machine(RVM) power quality. This part of the thesis is the sample collection, sample generation parameters are set with reference to the provisions of the power system parameter setting, the maximum to meet authenticity. Matlab simulation generated by the random sample for HHT can accurately extract the amplitude, frequency and phase information, the amplitude information calculation method applied voltage RMS acquisition phase standard deviation, the standard deviation of the frequency, for the preparation of these three quantities RVM, then sampled training and testing.RVM is a two-component classifier, so you want to apply a combination of multiple classifiers to achieve multi-class effect, so that to achieve the classification of complex samples, sample training conducted for the amount of sample characteristics to be classified, the training base of the larger sample classification effect better.Thesis RVM classification results for testing, to verify the accuracy of the classification, since samples are drawn from the matlab simulation, for the type and quantity of the sample are known, the RVM classification results have verified the data support. Then blend of power quality samples for automatic identification, data showed a high classification accuracy, is an effective classification method.