A method based on sparse component analysis-variational modal decomposition (SCA-VMD) was proposed to separate transformer operation sound signal and reduce noise. The under-determined blind source separation based on sparse characteristics can realize the effective separation of source signals when the number of observed signals is less than the number of unknown source signals

and the variational modal decomposition (VMD) can decompose a multi-component signal into multiple single-component signals at one time. The two-channel observation signals were used as input

the transformer operation acoustic signals were separated by SCA

the separated signals were decomposed into four-layer intrinsic mode function(IMF) components by VMD

the high frequency components and low frequency components were denoised by the method of threshold filtering

and the denoising signals were reconstructed by the new IMF components. The results of simulation and experiment show that this method can effectively separate and de-noising the operating sound signal of transformer.