基于小波神经网络的配电网故障类型识别  

Fault Type Identification in Distribution Network Based on Wavelet Neural Network

作　　者：赵智[1] 王艳松[2] 鲍兵[1] 赵中山[1] 陈学梅[1] 陈津刚[1] 

机构地区：[1]大港油田公司采油工艺研究院,天津300280 [2]中国石油大学(华东)信息与控制工程学院,东营257061

出　　处：《电力系统及其自动化学报》2007年第6期93-96,共4页Proceedings of the CSU-EPSA

摘　　要：为准确可靠地识别配电网故障类型,应用小波变换技术对故障信号进行预处理,滤除其中大量的谐波和非周期分量,准确地提取工频信息构成神经网络的训练样本集,通过构建小波神经网络实现配电网故障类型的识别。仿真测试表明,此网络模型收敛速度快,并能在各种故障模式下准确实现故障类型的识别,不受故障过渡电阻、系统运行方式以及故障点位置等随机因素的影响。In order to identifiy fault type in distribution network accurately,wavelet transform technique is applied to pretreat the fault signal to eliminate plenty of harmonic and aperiodic component. The power frequency information is extracted exactly to compose training sample of neural network. Distribution network fault type is identified by wavelet neuval network （WNN）. The simulation results show that the WNN model has fast convergence performance, it is able to identify the fault type accurately in various fault modes in the influence of the random factors such as fault transition resistance, system running mode and fault location.

关 键 词：故障类型识别 小波变换 人工神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]