The Application Of Wavelet Theory In The High-voltage Power Metering System Fault Detection

Electricity metering is non-substitutable in the normal operation of electricity system, not onlydoes it matter the production, research and management of electricity enterprises, but also the immediateinterests of widespread electricity users.The failure of the metering system, including components of themeasurement within the metering system and human-caused changing within the system wiring fault,themalfunction will be made by spontaneously or human-caused,no matter what form of failure will result ininaccurate of measurement system.It is possible that the incorrect wire-connection of three-phase ammeter,the disconnection ofvoltage transformer or current transformer,the one or two sides short-circuit of current transformer and thesecondary interphase with short circuit of current transformer,are the reasons that lead the malfunction ofelectricity metering system.The last two reasons, with the Polarity reverse connect fault,and the detectiontechnology of Watt-hour meter phase lack fault are the hot topic of electricity metering system currently.Detecting the current transformer short-circuit is a key technology of metering system malfunction. How todetect the malfunction is quite important, especially for current transformer earth faults. It has importantacademic and realistic meaning to find out the methods of detecting these faultsReferred to lots of literature, the paper sets the one-time measurement short circuit of currenttransformer as an example, uses the wavelet analysis to remove noise signals of metering system faults andextract the faults features, and diagnoses type of faults by BP neural network. The main works of this paperare as follows:(1)introducing the development,consistence,analysising the type of fault of electricity meteringsystem produced;and choosing the Db5wavelet as basic wavelet because of its relatively energyconcentration and local properties,then denoising the signal which extracted by metering system,1KHZsignal as excitation signal is joinedin the metering system,the type of fault is detected by1KHZ signalchanging, using the best wavelet packet basis to remove the noise of electricity metering system faults.Thede-noising process is divided into four parts,which are analysising signal by wavelet packet,de-noising bythe best wavelet packet,thresholding the wavelet packet coefficient,and reconstructing the signal by wavelet packet,the simulation shows that the result is obvious based on DB5optimal wavelet packet in the meteringsystem fault signal de-noising.(2)we extract feature component of the faults after singal be de-noisd;use the Db10to detect thesingular points of faults;use the energy conservation principle of decomposing the signals with waveletpacket and extracting the faults feature.Based on the characteristics of the extracted fault signal, wedetermine the fault type is current transformer short-circuit.