摘要
电力调度中心为进一步提高故障类型识别率和计算速度,利用提升小波和BP网络构造了新的小波神经网络故障识别模型,应用db5提升小波对故障电流进行分解,将分解到的(0,375)Hz频率段的系数输入到BP神经网络;为了提高算法的收敛速度,采用共轭梯度法训练该神经网络。通过ATP仿真及华东电网实际故障录波数据的测试,结果表明该模型具有很高的识别率和收敛速度。
To further improve the fault classification rate and calculation speed, a novel fault classification model using the lifting wavelet and BP network was developed. The coefficients of the fault current in the low frequency band between 0 and 375 Hz that decomposed by db5 lifting wavelet were put into the BP neural network, at the same time, the conjugate gradient method was adopted to train the network in order to improve the convergence speed of the algorithm. ATP simulation and tests of the real recording oscillograph data of fault occurred in East China Power Grid prove that the model has the advantages of high classification rate and convergence speed.
出处
《华东电力》
北大核心
2006年第2期29-33,共5页
East China Electric Power
关键词
故障诊断
故障类型识别
录波数据
BP网络
提升小波
共轭梯度法
fault diagnosis
fault classification
recording oscillograph data of fault
BP network
lifting wavelet
conjugate gradient method
作者简介
王忠民（1968-），男，硕士，主要研究方向为电力系统继电保护及其软件工程。
