Transient Power Quality Disturbance Detection Analysis

With the development of the national economy, energy occupies an increasinglyimportant position in today’s life, the composition of power grid changes in recentyears: on one hand, in the power supply side, the intermittent energy power gridcapacity increases, there will be some fluctuations in the power output and lead thedecline of the power quality; On the other hand, in the load side, the access ofnon-linear loads will lead the decline of the power quality. This will affect the normaloperation of sensitive electrical equipment and cause great economic losses.Consequently, power quality, especially the transient power quality has become a hotresearch topic.The paper analysis two kinds of power quality detection method based onmathematical transformation: Fourier transform and wavelet transform. Thencompares these two methods’ effect in signal de-noising. Simulation platform forPSCAD/EMTDC software, I build a wind power the access system simulation model,simulate several transient voltage disturbances easy caused by the wind power systemand collect the relevant disturbance data. Combining the advantages of these twokinds of detection methods, I proposed power quality detection method based on thecombination of Fourier transform and wavelet transform. Precede spectrum analysisfor disturbing signal with Fourier transformation, analyze the noisy part of the signaland de-noise with spectrum, breakdown de-noising signal with wavelet transform andlocate the starting and ending time of signal disturbance according to the waveletcoefficient.Identify the signal disturbance based on the combination of wavelet transformand BP neural network. Firstly, decompose the signal with wavelet transform; we canget different layers of wavelet coefficients. Then, input them to the energy functionwhich could identify a variety of transient disturbance and divide he processed data into two categories: training data and testing data. At last, train BP neural networkwith training data and test the trained neural network with testing data. Simulationdata comes from the data the wind farm collection bus access to system when thetransient disturbance occurs, it has practical significance.