SEPARATION OF SUBSTATION NOISE COMBINING SPECTRAL SUBTRACTION AND VARIATIONAL MODE DECOMPOSITION

In recent years, as China's urbanization process has accelerated, the load on the power supply has increased, and many substations have been built in cities where the population is concentrated. While the construction of substations in cities eases the pressure on urban power supply, it also causes serious noise pollution in urban areas. In order to effectively control the noise pollution from substations, it is necessary to accurately measure and evaluate the noise inside the substation and the surrounding areas. However, in urban conditions, there are traffic noise, environmental noise and other background noise interference, resulting in the sample noise contains multiple signals. Therefore, the separation of substation noise and background noise in the noise measurement process will be very conducive to making an accurate estimate of the propagation of substation noise. To address this problem, this paper proposes a substation noise extraction model that combines spectral subtraction and variational modal decomposition. The model performs distributed sampling at the main sound sources of the substation and in the surrounding noise-sensitive areas. Then, non-stationary noise is removed using spectral subtraction. Finally, the remaining signal is decomposed using the variational modal method. The decomposed signals are analysed for correlation with the sampled signals at the substation to determine the proportion of substation noise in the mixed signals for substation noise extraction. Experimental verifications show that this method can effectively extract the substation noise components in the sampled signals from noise-sensitive places. The method can provide effective support for substation noise control.