基于神经网络的电晕电流变化曲线拟合数学模型研究  

Study on mathematical model of corona current curve fitting based on neural network

作　　者：贺春[1] 刘大维 杨志华[1] 鲁轩 李宁[1] 徐锋 晏鹏兵 He Chun;Liu Dawei;Yang Zhihua;Lu Xuan;Li Ning;Xu Feng;Yan Pengbing(State Grid Tianjin Power Maintenance Company,Tianjin 300143,China;Zhengtaihe Energy Technology(Beijing)Co.,Ltd.,Beijing 100020,China)

机构地区：[1]国网天津市电力公司检修公司,天津300143 [2]政泰和能源科技(北京)有限公司,北京100020

出　　处：《国外电子测量技术》2021年第1期65-69,共5页Foreign Electronic Measurement Technology

基　　金：国网天津市电力公司2020年研究开发项目(KJ20-1-50)资助。

摘　　要：为了解决目前电晕电流曲线拟合误差大的难题,提出了基于神经网络的电晕电流变化曲线拟合数学模型。首先对电晕电流曲线拟合数学模型的研究进展进行分析,找到不同电晕电流曲线拟合数学模型的不足,然后通过BP神经网络强大逼近能力和非线性处理能力拟合存在双指数函数特征的电晕电流变化规律,最后进行了具体的电晕电流曲线拟合数学模型求解实例分析,结果表明,BP神经网络可以高精度拟合电晕电流曲线变化趋势,电晕电流曲线拟合误差小,通过参数阈值和权值得到电晕电流变化曲线的数学表达式,可以适用拟合各种电晕电流变化曲线。In order to solve the problem of large fitting error of corona current curve,a mathematical model of corona current curve fitting based on neural network is proposed.Firstly,the research progress of corona current curve fitting mathematical model is analyzed,and the shortcomings of different corona current curve fitting mathematical models are found.Then,the corona current variation law with double exponential function characteristics is fitted by the strong approximation ability and nonlinear processing ability of BP neural network.Finally,the specific mathematical model of corona current curve fitting is analyzed,the results show that the BP neural network can fit the corona current curve with high accuracy,and the fitting error of corona current curve is small.The mathematical expression of corona current curve is obtained by parameter threshold and weight,which can be used to fit various corona current curve.

关 键 词：神经网络 电晕 电流 变化曲线 拟合数学 阈值 

分 类 号：TN365[电子电信—物理电子学]