摘要
变压器的顶层油温会影响变压器的绝缘性能和运行寿命,精确预测顶层油温对提高变压器的利用率至关重要。针对目前顶层油温预测准确度不高的现状,提出一种提高顶层油温预测精度的组合模型,并以实际算例验证了模型的有效性。首先,利用Susa D热路模型预测顶层油温得到初始值;其次,建立BP神经网络模型预测热路模型的误差;最后,利用预测的误差结果修正热路模型的初始值。实例分析表明,组合模型较Susa D热路模型及单一预测模型预测精度更高。
The top-oil temperature of the transformer has an influence on its insulation performance and service life.It is essential to accurately predict the top-oil temperature for improving the utilization of the transformer.To improve the accuracy in prediction of top-oil temperature,this paper proposes a combination model with high accuracy and it is validated by the practical examples.First,Susa D's thermal circuit model is applied to predict the initial value of the top-oil temperature.Then,the BP neural network model is established to predict the errors.Finally,the initial values are amended using the predicted error results.The proposed model is demonstrated to be more accurate than Susa D's thermal circuit model as well as the single prediction model through case study.
作者
杨欢红
丁宇涛
宋亮
俞京锋
阮远峰
YANG Huan-hong;DING Yu-tao;SONG Liang;YU Jing-feng;RUAN Yuan-feng(Electric Power Engineering,Shanghai University of Electric of Power,Shanghai 200090,China;State Grid Zhejiang Deqing Power Supply Company,Deqing 313200,China)
出处
《水电能源科学》
北大核心
2018年第8期171-174,共4页
Water Resources and Power
关键词
变压器
顶层油温
组合模型
热路
BP神经网络
transformer
top oil temperature
combination model
thermal circuit
BP neural network