摘要
变压器热点温度是影响变压器运行状态、物理条件和绝缘寿命的重要因素之一。为了避免植入光纤测量热点温度时造成不良影响,本文通过推导传热学方程,提出了一种基于流线和支持向量回归机(SVR)的热点温度反演方法。以10kV单相变压器模型为例,进行了不同热源、对流换热系数和环境温度的温度场数值计算,并提取了三类主流线,研究了不同流线点数目、位置和不同流线根数下的热点温度反演结果,得出了选取同时流经热点和变压器外壳附近的流线时,热点温度反演精度最高的结论。此外,在众多该类流线中,仅选择一条流线并在该流线上选择温差最大的两个点所对应的外壳点作为温度测点时,热点温度反演误差最小,误差均在±0. 5℃以内。该方法实现了特征量测量非植入,比传统方法具有更高的精度,并为变压器状态实时监测、热点温度计算提供了一种新途径。
The transformer hot-spot temperature is one of the important factors influencing the running state,physical condition and insulation life of the transformer.In order to avoid the adverse effects of optical fiber implant on hot spot temperature,a method of hot-spot temperature inversion and streamline based on SVR and streamline is proposed through the analysis of the thermodynamic equations in oil circulating.Taking a 10kV single-phase transformer model as an example,the numerical temperature field calculation of different heat source,heat transfer coefficient and environmental temperature is carried out.Then,three kinds of main streamlines are extracted,and the inversion results of hot-spot temperature under different streamline points number and location,together with different streamline root number have been studied.It is concluded that choosing streamline which is through the hot spot and the transformer shell at the same time can get better inversion results.In addition,in a number of these streamlines,only choosing one streamline and selecting two points of the maximum difference in temperature can get better results.It can obtain the minimum error and the error was within 0.5 degrees by using the shell point corresponding streamline point to invert the hot spot temperature.This method realizes that the measurement of the characteristic quantity is completely non-implantable,has a higher precision than the traditional method,and provides a new approach for monitoring transformer condition constantly and predicting the hot spot temperature as well.
作者
全妤
阮江军
龚若涵
陈荆洲
金颀
文武
QUAN Yu;RUAN Jiang-jun;GONG Ruo-han;CHEN Jing-zhou;JIN Qi;WEN Wu(School of Electrical Engineering,Wuhan University,Wuhan 430072,China)
出处
《电工电能新技术》
CSCD
北大核心
2018年第11期23-31,共9页
Advanced Technology of Electrical Engineering and Energy
关键词
流线
支持向量回归机
非植入
热点温度
反演模型
streamline
support vector regression
non-implantable
hot temperature
inversion model
