基于数据增强技术的LSTM模型变压器故障诊断研究
Transformer Fault Diagnosis of LSTM Model Based on Data Enhancement Technology
蔡晨1
作者信息
- 1. 柳州铁道职业技术学院 继续教育学院, 广西 柳州 545616
- 折叠
摘要
为了解决变压器故障诊断中存在着依靠人工经验等方面的问题,提高变压器故障诊断的能力,笔者提出了基于数据增强的长短期记忆网络(long short-term memory,LSTM)模型的变压器故障诊断方法.首先通过数据增强技术增加数据样本量,然后利用LSTM构建变压器故障诊断模型,最后进行变压器故障诊断实验.结果表明:该方法预测的准确率、查准率、查全率及F1 值均达到 0.998;与支持向量机模型比较,各项评价指标至少提高 8%.该方法能够提高变压器故障诊断能力,有助于变压器故障的诊断与维修.
Abstract
In order to solve the problems of relying on manual experience in transformer fault diagnosis and improve the ability of transformer fault diagnosis,the author proposes a transformer fault diagnosis method based on data enhanced long short-term memory(LSTM)model.Firstly,the data sample size is increased by data enhancement technology,then the transformer fault diagnosis model is constructed by LSTM,and finally the transformer fault diagnosis experiment is carried out.The results show that the accuracy rate,precision rate,recall rate and F1 value of this method are 0.998;Compared with the support vector machine model,each evaluation index is increased by at least 8%.This method can improve the ability of transformer fault diagnosis and contribute to the diagnosis and maintenance of transformer faults.
关键词
变压器/故障诊断/数据增强/LSTM模型Key words
transformer/fault diagnosis/data enhancement/LSTM model引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据