基于广义回归神经网络的变压器表面振动基频幅值计算

发布时间：2018-01-07 16:33

本文关键词：基于广义回归神经网络的变压器表面振动基频幅值计算　出处：《高电压技术》2017年07期 　论文类型：期刊论文

【摘要】：变压器振动基频幅值大小是分析和评判变压器运行状态和故障诊断的重要依据。为此,提出一种基于广义回归神经网络(GRNN)的变压器振动基频幅值计算方法,用于计算正常运行状态下的变压器表面振动基频幅值。所提方法考虑变压器振动产生机理和影响因素,先根据变压器运行电压、负载电流、油温等历史运行工况数据以及表面振动历史数据进行GRNN训练,保存训练好的GRNN网络即可根据实时运行工况数据计算变压器表面振动基频幅值。某台在运变压器表面振动实测信号的计算结果表明:所提方法比现有方法计算误差大幅下降约50%。研究结果可为变压器振动在线监测提供重要参考。
[Abstract]:The amplitude of fundamental frequency of transformer vibration is an important basis for analyzing and judging the operating state and fault diagnosis of transformer. A method for calculating the fundamental frequency amplitude of transformer vibration based on generalized regression neural network (GRNN) is presented. The method is used to calculate the fundamental frequency amplitude of transformer surface vibration under normal operation. The proposed method considers the mechanism of transformer vibration and the influencing factors. Firstly, according to the transformer operating voltage and load current, the proposed method is used to calculate the fundamental frequency amplitude of the transformer surface vibration. Oil temperature and other historical operating condition data and surface vibration history data were trained by GRNN. The basic frequency amplitude of transformer surface vibration can be calculated according to the real-time operating condition data by saving the trained GRNN network. The calculation results of the measured signal of the surface vibration of a certain transformer show that:. The calculation error of the proposed method is about 50% lower than that of the existing methods. The research results can provide important reference for on-line monitoring of transformer vibration.
【作者单位】：华北电力大学电气与电子工程学院;
【基金】：中央高校基本科研业务费专项资金(2016MS96)~~
【分类号】：TM41;TP183
【正文快照】： 0引言1变压器是电力系统的关键设备之一,其运行状况对电网的安全、稳定运行具有重要作用。绕组和铁芯是变压器的重要部件且故障多发,而目前还没有简单、有效的在线异常检测和故障诊断方法。变压器表面振动与内部机械结构关系密切,国内外学者对基于振动分析的变压器在线监测和

【参考文献】