Wavelet-based techniques have been widely used to extract partial discharge (PD) signals from noisy signals. Generally, the procedure consists of 3 steps: wavelet selection, decomposition scale determination, and noise estimation. Wavelet selection is the first and most important step for its successful application in PD denoising. However, despite many variants of techniques deployed, the success rate is not generally good especially when the signal to noise ratio is unity or less. This paper discusses a novel technique that addresses this issue. The technique is inspired by the concept of Shannon entropy and the associated information cost functions (ICF) in information theory. It is adaptive to the detected PD signals. The paper demonstrates that the proposed technique is effective when applied to PD signals obtained through laboratory experiments and on-site measurements. When this technique is applied to cable diagnostics, it should have the potential to extend the range of PD detection from cables.
|Number of pages||19|
|Journal||CIGRE Science and Engineering|
|Publication status||Published - 30 Jun 2019|
- partial dishcarge
- wavelet selection
- wavelet entropy