摘要
针对电力电缆故障的关键因素在于电缆分接头的温升。本文提出采用总体平均经验模态分解EEMD和改进Elman神经网络的方法对非线性电力电缆接头温度时间序列进行预测的方式达到预警电力电缆故障的目的。首先,采用光纤光栅传感器来监测电缆接头温度,得到样本序列。为避免模态混叠,采用EEMD对电力电缆接头温度序列进行预处理。预处理后的温度序列为一系列彼此独立的本征模函数分量,外加一个残余分量。为提高Elman神经网络的预测精度,在预测前利用相空间重构法对分解后的序列进行重构计算。在对重构序列数据进行归一化后利用Elman神经网络进行下一周期内的温度预测。实际预测结果表明,本文算法具有较高的拟合精度,适应性较好,可以推广应用至实际电力电缆故障的监测和预警系统。
The key factor of power cable fault is the temperature rise of cable tap.In this paper,the method of average ensemble empirical mode decomposition(EEMD)and the improved Elman neural network are used to predict the nonlinear temperature and time series of power cable joints to achieve the early warning purpose of power cable faults.First,a fiber grating sensor was used to monitor the temperature of the cable joint to obtain a sample sequence.To avoid modal aliasing,EEMD is used to preprocess the temperature sequence of power cable connector.The temperature sequence after pretreatment is a series of independent eigenmode function components plus a residual component.In order to improve the prediction accuracy of Elman neural network,the reconstruction method of phase space is used to reconstruct the decomposed sequence before prediction After normalization of reconstructed sequence data,Elman neural network was used to predict the temperature in the next period.The actual prediction results show that the proposed algorithm has high fitting precision and good adaptability,which can be widely used in the monitoring and early warning system of actual power cable faults.
作者
杨涛
白晗
刘洋
王云龙
张楚曼
YANG Tao;BAI Han;LIU Yang;WANG Yunlong;ZHANG Chuman(State Grid ShenYang Power Supply Company Cable Transport Inspection Office,Liaoning,Shenyang,China 110031)
出处
《实验室研究与探索》
CAS
北大核心
2018年第10期59-62,共4页
Research and Exploration In Laboratory
关键词
电力电缆
接头温度
总体平均经验模态分解
ELMAN神经网络
光纤光栅
电缆故障预判
power cable
joint temperature
average ensemble empirical mode decomposition
Elman neural network
fiber optic grating
cable fault pre-judgement
作者简介
杨涛(1974-),男,辽宁沈阳人,本科,助理工程师,输配电运行、工程方面研究。Tel.:13940300525;E-mail:syauhangang@163.com。
