摘要
为了正确识别变压器励磁涌流和短路电流,提出一种基于经验模态分解(Empirical mode decomposition,EMD)和支持矢量机(Support vector machine,SVM)的识别方法。该方法首先对原始电流信号进行经验模态分解,将不平稳信号分解为多个平稳的固有模态函数(Intrinsic mode function,IMF)之和,分别计算前五层IMF分量能量并组成能量特征向量;然后以此作为SVM分类器的输入参数来识别励磁涌流和短路电流。仿真结果表明,该识别方法在小样本情况下,能准确、有效地识别励磁涌流和短路电流两类电流信号,而且受噪声的影响小。
In order to accurately identify inrush current and the method based on empirical mode decomposition （EMD） short circuit current of transformer, this paper proposes and support vector machine （SVM）. Firstly it conducts empirical mode decomposition of the original current signal, decomposing the unstable signal into several stable intrinsic mode functions （IMF）. Secondly, it calculates the component energy of the former five level of IMF. Finally, taking the calculation result as input parameter of SVM, it identifies inrush current and short circuit current. The simulation shows that this method, under the condition of small sample, is able to accurately and efficiently i-dentify inrush current and short circuit current with low influence of noise.
出处
《黑龙江电力》
CAS
2013年第4期303-306,共4页
Heilongjiang Electric Power
关键词
励磁涌流
经验模态分解
支持矢量机
特征向量
inrush current
empirical mode decomposition
support vector machine
eigenvector
