Study on Reconstruction of Transformer Factory Noise Signal

Yuchao Ma,Yong Qian,Juan Mo,Li Han,Xin Wang
DOI: https://doi.org/10.1117/12.3015437
2024-01-01
Abstract:In this paper, a study is made on the distribution of noise spectrum characteristics of a 500 kV transformer operating at 80% and 100% of the rated current and 80% and 100% of the rated voltage. To improve the ability of identifying transformer noise in different factory conditions, the short-time strong correlation of line spectrum of transformer vibration noise is used to obtain high-quality reconstruction signal, the line spectrum SNR over detection threshold is extracted through CFAR processing, and 100 Hz and frequency doubling are identified. The test results show that the reconstruction signal is improved by 3–5 dB compared with the detected SNR of original line spectrum, and more vibration line spectrum can be extracted, thus providing more information about the line spectrum for diagnosis, monitoring and identification of transformer factory noise fault.
What problem does this paper attempt to address?