Transformer Excitation Inrush Current Based On Neural Network Identification Method Of The Research

with the rapid development of national economy, electric power system’s scale expandsunceasingly, for the safety and reliability of the power equipment put forward higher request.Power transformer is the most important electrical equipment in power system, so in view ofthe transformer protection is particularly important. Longitudinal differential protection haslong been widely used as main protection of transformer, but the existing transformerdifferential protection has long been beset by unbalance current, especially the unbalancecurrent of excitation inrush current. So how to distinguish transformer excitation surge andinternal fault current is a core problem to enhance the reliability of differential protection. Butcurrently used in differential protection of excitation inrush current identification methods arenot well meet the requirements of transformer protection, therefore it is necessary to explorethe recognition speed is faster, more accurate new method.Paper first analyzes the cause of excitation inrush current of transformer and its influencefactors, and the impact on the transformer differential protection, and for a variety ofexcitation inrush current identification methods were analyzed and evaluated. In this paper,using Simulink on the no-load switching-in of excitation inrush current are simulated, and onthis basis, the characteristic has carried on the thorough analysis of the inrush currentwaveform. At the same time, in order to analyze the influencing factors of flow, the paperrespectively change the power supply early closing phase Angle and transformer iron coreremanence, and compared the simulation excitation inrush current of transformer, theobservations at the beginning of the closing phase Angle and the residual magnetism on theinfluence of inrush current waveform. Further, in order to compare excitation inrush currentand internal fault current waveform, simulation, this paper discusses the internal fault oftransformer by the Powergui module of Simulink bring excitation inrush current and faultcurrent harmonic analysis, statistics of their higher harmonic content. Paper finally collectedwith FFT analysis of excitation inrush current and internal fault current2~5times offundamental wave and harmonic of LVQ network input, the LVQ network to do a lot oftraining and validation of the simulation, the results show that the LVQ neural network canidentify accurately and rapidly excitation inrush current.This article first proposed the methods of identify transformer excitation inrush currentbased on LVQ neural network. The paper will LVQ network and BP network were analyzed,respectively, from the identification speed, precision and accuracy of the three parties facetwo network to do the simulation analysis, results show that the LVQ network can more quickly and accurately identify transformer excitation inrush current. The scheme proposed inthis paper combined with transformer microcomputer protection using can improve thereliability of differential protection, LVQ neural network has broad application prospects intransformer protection.