摘要
首先,考虑到气体浓度序列的波动性,利用经验模态分解(Empiricalmode decomposition,EMD)方法,将油中溶解气体浓度序列分解为不同特征尺度的本征模态函数(Intrinsic mode function,IMF)分量和1个剩余分量;然后,采用多粒度级联森林(Multi-grained cascadeforest,gcForest)模型对分解得到各子序列分量分别进行预测;最后,叠加所有各子序列分量的预测值作为最终结果。算例分析结果表明,相较传统预测方法,所提的EMD-gcForest方法具有较高的预测精度和泛化能力。
Firstly,considering the volatility of gas concentration series,the empirical mode decomposition(EMD)method is used to decompose the dissolved gas concentration series in oil into intrinsic mode function(IMF)components with different characteristic scales and one residual component;than,a multi granular cascade forest(gcForest)model is used to predict the components of the decomposed subsequences;finally,the predicted values of all subsequence components are superimposed as the final result.The analysis results of numerical examples show that the proposed EMD-gcForest method has higher prediction accuracy and generalization ability compared to traditional prediction methods.
作者
张鹏坤
余进
李波
单长吉
张靖
ZHANG Pengkun;YU Jin;LI Bo;SHAN Changji;ZHANG Jing(College of Physics and Information Engineering,Zhaotong University,Zhaotong 657000,China)
出处
《电力科学与工程》
2023年第6期32-38,共7页
Electric Power Science and Engineering
基金
云南省科技厅地方本科高校联合专项资金(112031401094)。
关键词
电力变压器
故障诊断
经验模态分解
多粒度级联森林
油中溶解气体
transformer
fault diagnosis
empirical mode decomposition
multi-grained cascade forest
dissolved gas in oil
