Based On Recursive Mode Decomposition And Quantitative Analysis Of Power Quality Disturbance Signal Analysis And Recognition

With the rapid development of national economy and seientific technology, a great deal of nonlinear, impact and unsymmetrical load increased signifieantly in the power system, power quality problems caused more and more seriouse eonomiclosses, how to effeetively reduce the impact of power quality problems has become a research focus of the technology workers. The first step to improve the power quality is the detection, analysis and identification of power quality disturbances. Only to detect power quality problems rapidly and aecurately, conduct effectively analysis, and identify the type of disturbances, power quality problems can be controled and governed effectively.This paper analysed power quality disturbance signals without noise and power quality disturbance signal contain30dB noise, and comparised the differents on recursion figure’s identification results between the power quality signal with noise and without noise, used NMF method to extracted features and then used neural network to the intelligent identification. The recurrence plot and the reconstruct trajectory of power quality disturbance signal based on phase space reconstruction intuitive shows that different characteristics between the different disturbance signals. Recurrence plots of disturbance signals without noise are smoother, they can reflect the obvious characteristics, but recurrence plots of signals with noise appear complex caused by the effect of noise, lines and graphics has great changs. However, in the figures, the disturbance characteristics are not affected, the recognition effect is very obvious.EMD method was used to decomposition the PQ disturbance signal, and the recurrence quantification analysis was first used to analyze the decopsed signal in this paper, and used this method into disturbance detection, recognition and location. By using EMD methods denoised six types signals of power quality disturbance contain the30dB noise (the interruption, voltage dip voltage rises, the voltage sag, transient oscillation, voltage spikes and transient harmonic), the noise in the high frequency part was removed and componented the low frequency signals, and the results show that noise can be removed from the original signal by using EMD denoising method, while retaining disturbance characteristic features in low frequency signals. The recurrence plot and the reconstruct trajectory of denoising signals showed obvious difference, and recurrence quantification analysis was used into the next test for completeing intelligent classification and disturbance location with the method of Neural Network, the result proves that the method for power quality problems of the research is very effective.Finally, EEMD method was used in the decomposition of power quality signals. In this paper, the signal decomposition and feature extraction base on EEMD method was researched on the basis of all kinds of power quality disturbance signal with noise, and analyzed the recurrence plot and the reconstruct trajectory of power quality disturbance of the extracted feature. The method of feature extraction based on EEMD is more superior than the EMD method on mode mixing and end effection, EEMD method is better to be used into the research of power quality disturbance, especially, the complex disturbance signal can be decomposed into some signals with different disturbance type.It is effective to introduce recurrence plot and recurrence quantification analysis into the research and processing of non-stationary power quality disturbance signals. The two dimension phase space diagrams and recurrence plot demonstrate the recurrence phenomenon of inner dynamic mechanism of PQ disturbance signal in different modes. Recurrence quantification parameters present the recurrence characteristics of power disturbance signal. Power disturbance modes can be differentiated with using the parameters extracted through recurrence quantification analysis and BP neural network. The reconstruction rate of recurrence analysis can be used to position the start and end point of disturbance.From the simulation results, it can be found that recurrence quantification analysis is a novel and direct inspection method of power disturbance mode inspection. It is also the basis of subsequent research upon PQ disturbance signal and the overlapping of various disturbance signals.