基于改进变分模态分解和小波阈值法的单相接地故障电流降噪

针对电力系统输配电线路发生单相接地故障时，电气设备间的电磁环境复杂，现场环境干扰严重导致故障录波装置采集到的故障零序电流信号含有大量噪声，影响后续选线准确率的问题，提出了一种改进VMD和小波阈值法联合的单相接地故障的零序电流降噪方法，通过北方苍鹰优化算法优化改进变分模态分解（VMD）对零序电流信号分解，引入自适应相关阈值对分解后的分量进行筛选，对噪声分量进行小波阈值法降噪，最后将信号进行重构。通过搭建模型进行仿真实验，所提算法比传统VMD降噪算法信噪比提高了5.52%~35.99%，均方根误差降低了12.78%~30.88%，与小波阈值降噪方法、EEMD-小波阈值降噪方法、CEEMDAN-小波阈值降噪方法相比，也都有明显的优势，并且在标准测试信号Heavy Sine信号和Bumps信号中进行实验验证了算法的适用性。

Denoising of Single-phase Grounding Fault Current Based on Improved VMD and Wavelet Threshold

When a single-phase grounding fault occurs in power system transmission and distribution lines, the electromagnetic environment between electrical equipment is complex, and the severe field environmental interference causes a lot of noise in the fault zero-sequence current signal collected by the fault recorder, which affects the accuracy of subsequent line selection. A denoising method for single-phase grounding fault zero sequence current based on improved VMD and wavelet threshold method is proposed. The northern goshawk optimization algorithm optimizes the modified variational mode decomposition (VMD) to decompose the zero-sequence current signal. The adaptive correlation threshold is introduced to filter the decomposed components, and the wavelet threshold method is used to denoise the noise components. Finally, the signal is reconstructed. By building a model to conduct simulation experiments, the proposed algorithm improves the signal-to-noise ratio by 5.52%~35.99% and reduces the root mean square error by 12.78%~30.88% compared with the traditional VMD denoising algorithm. Compared with the wavelet threshold denoising method, EEMD wavelet threshold denoising method, and CEEMDAN wavelet threshold denoising method, the proposed algorithm also has obvious advantages, and the algorithm''s applicability is verified by experiments in standard test signals Heavy Sine signal and Bumps signal.