基于GRNN-MC的变压器振动信号预测  

Vibration signals prediction of power transformer based on GRNN-MC

作　　者：钱国超 王山 张家顺 代维菊 朱龙昌 王丰华[3] QIAN Guochao;WANG Shan;ZHANG Jiashun;DAI Weiju;ZHU Longchang;WANG Fenghua(Electric Power Research Institute,Yunnan Electric Power Company,Kunming 650217,China;Nujiang Power Supply Company of Yunnan Power Grid Co.,Ltd.,Nujiang 673100,China;Department of Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]云南电网公司电力科学研究院,云南昆明650217 [2]云南电网有限责任公司怒江供电局,云南怒江673100 [3]上海交通大学电气工程系,上海200240

出　　处：《电工电能新技术》2024年第3期41-48,共8页Advanced Technology of Electrical Engineering and Energy

基　　金：云南电网公司科技项目(YNKJXM20210025)。

摘　　要：变压器振动信号是评估其工作状态的重要参数之一,与绕组松动或变形等隐患密切相关,为揭示变压器振动信号的变化趋势,本文提出了一种基于广义回归神经网络和马尔科夫链修正的变压器振动信号预测方法。即分别以变压器运行电压、负载电流和振动信号归一化特征频率为输入和输出建立变压器振动信号广义回归神经网络预测模型,然后引入马尔科夫链并结合负载电流的变化对振动信号计算结果进行修正以获得最终的预测结果。对某500 kV变压器振动在线监测信号的分析结果表明:经马尔科夫链修正后的变压器广义回归神经网络振动信号预测模型预测精度高,可为变压器绕组运行状态的振动监测技术提供重要参考。Vibration signals of an operating transformer has becoming one of the parameters for the evaluation of the working condition of the transformer.To further investigate the variations of vibration signals of a power transformer,a method based on generalized regression neural network(GRNN)and Markov chain correction is proposed in this paper to predict the vibration signals of transformer.The input and output of the established GRNN are the operating voltage,load current and the normalized feature frequency of vibration signals.Then the Markov chain is used to correct the relative error between the calculation results of vibration signal model and the measured vibration signals.The prediction results of vibration signal of transformer are finally obtained.The calculated results of the on-line monitoring vibration signals of a 500 kV transformer show that the vibration signal prediction model of transformer by the generalized regression neural network model and the Markov chain correction is capable of predicting the vibration signals of transformer with higher accuracy.The results can provide an important reference for vibration monitoring technology of power transformer.

关 键 词：变压器 振动信号 广义回归神经网络 马尔科夫链 归一化特征频率 

分 类 号：TM411[电气工程—电器]