An Automatic Tool for Partial Discharge De-Noising via Short-Time Fourier Transform and Matrix Factorization

This paper develops a fully automatic tool for the denoising of partial discharge (PD) signals occurring in electrical power networks and recorded on-site measurements. The proposed method is based on spectral decomposition PD measured signal via joint application short-time Fourier transform singular value decomposition. estimated noiseless reconstructed clever selection dominant contributions, which allows us to filter out different spurious components, including white noise discrete spectrum noise. offers viable solution can be easily integrated within measurement apparatus, with unavoidable beneficial effects detection important parameters localization. performance first demonstrated synthetic test then it applied real data. A cross comparison other state-of-the-art alternatives included study.