摘要
针对电力系统配电线路故障类型识别的问题,为提高故障类型识别准确性,提出应用小波变换技术对故障信号进行预处理,提取工频信息构成神经网络的训练样本集,通过构建自组织特征映射网络对不同故障类型的特征向量进行自动聚类来实现对故障类型的识别。大量的仿真测试表明,此网络模型收敛速度快,通过自学习能够有效覆盖故障模式空间,实现对不同故障类型的准确识别,网络对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因素的影响。
To the problem of identification of distribution line in power system, In order to improve the accuracy of fault identification, wavelet transformation technique is used to pretreat the fault signal, eliminate plenty of harmonics and most aperiodic component, extract fundamental information exactly, it is used as train-set of neural network. Realizing distribution network fault type identification by constructing Self-Organizing Feature Map neural network and processing automatic clustering to different character vectors related to fault type. The simulation analysis result indicates that the SOM model has fast convergence performance, it can effectively identify different fault type. It is able to identify the fault type accurately in various fault models in the influence of the random factors such as fault resistance, system running mode and fault place.
出处
《自动化技术与应用》
2012年第9期64-68,共5页
Techniques of Automation and Applications
关键词
配电网
故障类型识别
小波分析
神经网络
distribution network
fault type identification
wavelet analysis
artificial neural network
