The Application And Research On Fault Diagnosis Of High Voltage Electric Power Measuring System Based On Wavelet Transform And Neural Network

With the rapid development of power system, high voltage electric power metering system has beenwidely used, and the security and reliability of the system have a great influence on the country’s economy.So the fault diagnosis technology of high voltage power system has been the research focus of high voltageelectric power metering system. Electrical energy has been widely used in various departments, can beeasily converted into other forms of energy, and can be suitable for long distance transmission, automaticcontrol, mass production and centralized management. Besides, it has obvious advantages in use. So theinfluence of interruption or reduction in the supply of electrical energy on national economy will beserious.Faults in high voltage electric power metering system mainly include the fault of the internalelements of the system and Man-made Fault. While the security of power system operation keeps beingimproved, Being able to realize real time on-line monitoring of electric power metering system and thejudgment of the initial stage of fault occurrence in electric power metering system has become a pressingproblem.According to the characteristics of high voltage electric power metering system faults and based on theprevious research, This paper applies wavelet transform theory and neural network theory to the faultdiagnosis of high voltage power system and does the following analysis and research.Firstly, the paper establishes the system model of primary side fault and second side fault of electricpower metering system and gives an analysis of a variety of faults. According to several representative fault types, the simulation is carried out with the software Multisim10. The test signals got from test points undervarious fault conditions prove that theoretical analysis is correct.Secondly, aiming at the short circuit fault of primary side of high voltage electric power meteringsystem, we designed the embedded real-time monitoring device of electric power metering system to putfault diagnosis theory into practice. We have given an analysis of the choice of microcontroller unit andcurrent transformer used in detecting circuit and made a study of envelope phenomenon and sharp pulsephenomenon.Finally, by using the improved soft-threshold algorithm, we make denoise treatment for fault signalsextracted from the detecting device of high voltage electric power metering system, then make use of dmeywavelet with excellent cutoff frequency performance to carry out a four-level wavelet decomposition offault signals, and use the data of the frequency band including the detecting signal to reproduce thewaveform and to study it. At last, we collect samples to train Elman network and use the advantage of goodself-adaptation of Elman neural network to apply it to the fault diagnosis of high voltage electric powermetering system. The results of simulation and test show that the method proposed by the paper caneffectively diagnose the fault of high voltage electric power metering system...