Study on Suppressing White Noises and Periodic Narrow Bandwidth Noise of PD by Vertically Invariant Gray-Level Morphological Filter

In this paper, a novel adaptive mathematical morphological filter is used to remove the noise, such as pulse noise, white noise, narrow bandwidth noise and so forth, in partial discharge signals. The morphological filter consists of vertically invariant morphological opening and closing operators. The shape of structuring element selected in this filter is disk. The filtering scales are varying according to different time positions to remove noises in partial discharge while preserving important features of signal because the signal or noise models are usually non-stationary in partial discharge on-line monitoring. Simulates and experiments are performed to show the availability and efficiency of the method introduced in this paper.