微型电脑应用2025，Vol.41Issue(2)：171-173,3.
基于类重叠分层随机森林的油中溶解气预测和长短期故障分析方法
Dissolved Gas in Oil Prediction and Long Short-term Fault Analysis Method Based on Class Overlapping Stratified Random Forest
徐欢 1杨秋勇 1杜浩文 2方永瑶3
作者信息
- 1. 南方电网有限公司,数字化部,广东,广州 510000
- 2. 南方电网数字传媒科技有限公司,广东,广州 510000
- 3. 南方电网大数据服务有限公司,广东,广州 510000
- 折叠
摘要
Abstract
In order to improve the efficiency of transformer fault prediction,a method for predicting dissolved gas in oil and ana-lyzing long short-term fault based on class overlapping stratified random forest is proposed.By monitoring the concentration of hydrogen,methane,ethane,acetylene and other gases in the transformer insulating oil,the historical time series data of three days before of the oil chromatogram label are used to train data,and the concentration of each gas in the transformer oil on the day after the label is predicted based on data modeling.The random forest model is built to establish a fault prediction model.By evaluating whether the prediction result of acetylene exceeds its set threshold,the possible fault of the monitoring device can be detected.The experimental results show that the overall prediction accuracy of the proposed model is as high as 99％,and the F1-score reaches 96％,which has better prediction and analysis effect.关键词
油色谱/变压器故障预测/切分变量/决策树/随机森林Key words
oil chromatogram/transformer fault prediction/cut-off variable/decision tree/random forest分类
电子信息工程引用本文复制引用
徐欢,杨秋勇,杜浩文,方永瑶..基于类重叠分层随机森林的油中溶解气预测和长短期故障分析方法[J].微型电脑应用,2025,41(2):171-173,3.
