Denoising method for transformer acoustic signal based on CEEMDAN-SVD-wavelet threshold
Xuchun Hou, Yong Qian, Yulong Che · Journal of Physics Conference Series · 2025
Abstract To address the interference of ambient substation noise on the collected transformer sound signals and enhance the quality and reliability of the signals, this paper proposes a hybrid denoising method that integrates CEEMDAN, SVD, and wavelet thresholding. Firstly, the transformer sound signals are decomposed using CEEMDAN to obtain a series of Intrinsic Mode Functions (IMFs), screening the IMF components based on the differences in kurtosis values between the inherent transformer noise and the ambient noise. Then, Singular Value Decomposition (SVD) is employed to extract the principal components and remove noise from each Intrinsic Mode Function (IMF). Subsequently, wavelet thresholding is used to further denoise the signal. Finally, the denoised signal is reconstructed. This paper uses two evaluation indicators, the signal energy ratio and the noise suppression ratio, to analyze the denoising effect, the results show that the method used in this paper effectively removes noise while preserving the fidelity of the useful signal, this is of great significance for improving the subsequent feature extraction and modal identification based on the denoised signals.