结合数学形态学滤波与频谱校正的发电机局部放电离散谱干扰抑制方法

A Method for Suppression Discrete Spectrum Interference in Partial Discharge of Generators Combining Mathematical Morphology Filter and Spectrum Correction

针对发电机局部放电在线测量中面临的脉冲信号被相似频率离散谱干扰淹没、现有方法难以快速准确提取的难题,提出一种结合数学形态学滤波与频谱校正的离散谱干扰抑制方法。首先利用形态学滤波器对原始信号频谱进行处理,提取脉冲信号的山丘状频谱,频域消减得到噪声频谱,进而基于经典阈值识别干扰频率粗值,然后通过频谱校正估计干扰参数并重构干扰波形,最后将重构波形从原始信号中消减得到脉冲信号。仿真与现场数据验证了该方法的有效性和优越性,与其他方法相比,该方法不需要进行复杂的小波分解与重构以及傅里叶逆变换,降低了结构元素确定难度,计算效率高,且基本不对脉冲信号的波形产生影响。

For the problem that pulses are overwhelmed by discrete spectrum interference(DSI) posing similar frequency, and thus are difficult to accurately extract in real-time in generator partial discharge online measurement, a new method for DSI suppression combining mathematical morphology filter(MMF) and spectrum correction was proposed. The proposed method employed MMF to filter spectrum of original signal and to release spectrum of DSI. After that, the frequencies of DSI were identified with the classical threshold criterion, which were corrected to reconstruct the waveform of DSI through spectrum correction. At last, the pulses were extracted by subtracting DSI from original signal. The simulation and field results verify the effectiveness and superiority of the proposed method over other methods, since the method avoids the wavelet decomposition, reconstruction and IFFT, and simplifies structure element determination, whilst not distorts pulse waveform as well.