摘要
针对电力系统输电线路一系列的短路故障,把希尔伯特黄变换(HHT)完成对暂态信号特征量的提取与概率神经网络(PNN)作为诊断故障分类器相结合的方式,作为对输电线路故障分类的方法。利用HHT能够充分反映局部暂态信号的特点,对集合经验模态分解(EEMD)后的故障信号进行小波阈值降噪,运用HHT进行重构,得到三相电流以及零序电流4组特征能量函数值,并作为4组特征量输入到经遗传算法优化过平滑因子的PNN中进行训练,最终得到分类器。经Matlab仿真实验显示,该方法能够有效优化信号波形并提高故障分类精度。
In order to eliminate a series of short circuit faults in transmission lines of power system, a way that combining the Hilbert-Huang transform（ HHT） of transient signal characteristic extraction and probabilistic neural network（ PNN） as a fault classifier was used to classiy the faults for transmission lines. HHT can fully reflect the characteristics of partial transient signals, the fault signal after the ensemble empirical mode decomposition（EEMD） is de-noised by wavelet threshold. By using the HHT refactoring, four groups feature energy function values of three phase current and zero sequence current are obtained, and used as four groups characteristic quantities to be input to the PNN whose smoothing factor is optimized by the genetic algorithm for training, finally the classifier is obtained. The Matlab simulation experiments show that the method can effectively optimize the signal waveform and improve the fault classification accuracy.
出处
《测控技术》
CSCD
2018年第1期19-22,26,共5页
Measurement & Control Technology
基金
辽宁省重点实验室项目(LJZS003)
关键词
线路故障分类
希尔伯特黄变换
概率神经网络
小波阈值降噪
line fault classification
Hilbert-Huang transform
probabilistic neural network
wavelet threshold de-noising
