摘要
油浸式电力变压器的运行寿命及负载能力与绕组热点温度密切相关。精确预测变压器绕组的热点温度,是有效预防变压器热故障、准确预测变压器运行寿命和优化变压器设计的关键技术之一。论文研究了绕组热点温度支持向量机建模。为提高模型预测的精确度,选用径向基核函数优化模型结构;利用遗传算法对参数进行寻优。结合实验室模拟温升变压器绕组温度实测数据,提取输入和输出的特征量,并划分训练集和预测集,建立了基于遗传优化支持向量机的变压器绕组热点温度预测模型。实验表明:应用本文模型预测结果与实测值基本一致,优于BP神经网络以及Elman神经网络的预测结果。
The operation life and the load capacity of the oil-immersed power transformers are closely related with the winding hot-spot temperature（HST）. Accurate prediction of the transformer winding hot-spot temperature is one of the key technologies of effectively preventing the thermal fault, accurately predicting the operation life of the transformer and optimizing design of transformer. This paper studies the support vector machine（SVM） modeling of the winding hot-spot temperature. In order to improve the accuracy of model predictions, and the RBF kernel function is selected to optimize the model structure; and optimized the parameters by the genetic algorithm（GA）. Combined with the measured temperature of the temperature-rise transformer windings simulated in the laboratory, the characteristic quantities are employed as inputs, and the HST is used as output of the SVM model, and the measured temperature are divided into training set and prediction set. The transformer winding hot-spot temperature prediction model of support vector machine optimized by genetic algorithm is built. The experiment show that, the prediction results of the GA-SVM model are basically identical with the measured temperature, and are better than the prediction results of BP neural network and Elman neural network.
出处
《电工技术学报》
EI
CSCD
北大核心
2014年第1期44-51,共8页
Transactions of China Electrotechnical Society
基金
国家创新研究群体基金(51021005)
国家重点基础研究发展计划(973)(2012CB215205)资助项目
关键词
油浸式变压器
绕组热点温度
支持向量机
遗传算法
Oil-immersed transformer, winding hot-spot temperature, support vector machine,genetic algorithm
