摘要
针对变压器油中溶解气体浓度预测中存在的变量取值范围影响预测精度问题,提出了基于核目标度量规则(Kernel Target Alignment,KTA)和支持向量机(Support Vector Machines,SVM)的油中溶解气体浓度预测方法。在分析油中溶解气体产生机理的基础上选取输入变量,采用KTA对输入变量进行尺度缩放来避免变量的取值范围影响SVM泛化性能问题,利用交叉验证法选择SVM的参数,建立油中溶解气体浓度的KTA-SVM预测模型。将所提出的方法与SVM和灰色模型进行比较,均方根误差分别为0.156 8、0.179 1、0.220 5,实验结果表明了所提出的方法具有较优的预测精度和泛化性能。
A new prediction method of dissolved gas content in transformer oil based on kernel target alignment （KTA） and support vector machines （SVM） is proposed to deal with the problem that the prediction precision is affected by the range of input variables. Firstly, input variables are selected by analyzing the generation mechanism of dissolved gases in oil, then, input variables are rescaled by the KTA feature rescaling method to avoid the problem that the generalization ability of SVM is influenced by the value range of variables; Secondly, the cross validation method is adopted to select the parameters of SVM; At last, KTA-SVM is used to forecast dissolved gas content in transformer oil. Utilizing root mean squares error indexes to analyze the performance of the proposed approach, SVM and grey model, root mean squares error is 0.156 8, 0.179 1, 0.220 5 respectively, experimental results show that the proposed prediction method has better prediction and generalization performance.
出处
《控制工程》
CSCD
北大核心
2017年第11期2263-2267,共5页
Control Engineering of China
基金
国家自然科学基金(51366013)
江西省教育厅科技项目(GJJ161015)
关键词
变压器油
油中溶解气体
预测
支持向量机
核目标度量
Transformer oil
dissolved gas in oil
prediction
support vector machine
kernel target alignment
