基于门控循环单元神经网络的配电网故障选线

Fault Line Selection of Distribution Network Based on Gated Recurrent Unit Neural Network

针对小电流接地系统发生单相接地故障时选线准确率低的问题,提出一种基于集成经验模态分解和灰狼算法优化门控循环单元神经网络超参数的故障选线方法。利用集成经验模态分解的方法对线路零序电流进行分解,提取本征模态分量并计算分量的能量,以本征模态能量值作为特征向量输入至优化后的门控循环单元神经网络中实现故障选线。经过大量仿真实验验证,该算法在不同故障类型的情况下均能准确、可靠地进行故障选线,并在数据丢失及噪声干扰等情况下依然能有效识别故障线路,具有较好的抗噪性、准确性和可靠性。

Ina small current grounding system,the accuracy of fault line selection is lowunder single-phase grounding fault.To solve this problem,a fault line selection method based on ensemble empirical mode decomposition(EEMD)and grey wolf optimizer optimized gated recurrent unit(GGRU)neural network hyper-parameters is proposed.EEMD is used to decompose the zero-sequence current of transmission line,the eigenmode component is extracted,and the component energy is calculated.The value of eigenmode energy is taken as an eigenvector and further input into the GGRU neural network,thus realizing the fault line selection.The results of a lot of simulations show that the proposed algorithm can select the fault line accurately and reliably under different fault conditions.Even in the cases such as data loss and noise interference,it can still effectively identify faulty lines,showing satisfying noise immunity,accuracy and reliability.