Locating and reconstructing transformer low-frequency noises with a 3D, six-microphone array

Abstract

This paper presents diagnosis and analyses of the sound fields radiated from power transformers in 3D space under normal operational conditions. Transformers are crucial components in the power system. Good condition of a transformer is an important factor to ensure safe and reliable operations of the entire power system. Diagnosis and analyses of transformer noise are challenging because of the complexity of the acoustic environment surrounding the power grid. Previous studies have revealed that transformer noise is predominantly concentrated in the low frequency range. This low-frequency nature of transformer noise has made it extremely difficult to pinpoint the precise source locations. The present study shows that by using the Sound Viewer system, which is built on the principles of passive SODAR (Sonic Detection And Ranging) and HELS (Helmholtz Equation Least Squares) methods, we can not only pinpoint the precise locations of noise sources of a transformer, but quantify individual source strengths. Specifically, SODAR enables one to locate multiple sound sources simultaneously in 3D space over the frequency range of 20–20,000 Hz, and the HELS method enables one to reconstruct the acoustic field and acquire the optimal approximation of the acoustic pressure distribution in 3D space, time-averaged acoustic intensities, and time-averaged acoustic powers of the individual acoustic sources. The accuracy in reconstruction depends on the SNR (Signal to Noise Ratio) of the input data. The higher the SNR is, the more accurate the reconstruction becomes. Moreover, by using a spatial filter, we can eliminate the interferences of unwanted sound sources and extract the time-averaged acoustic power of a specific target. This salient feature enables us to perform a source ranking, which can be critical in designing the most cost-effective noise mitigation strategy. Results of the present study demonstrate that this technology can play a significant role in diagnosing and analyzing complex acoustic field in a non-ideal test environment, especially for transformer health monitoring and predictive maintenance in power systems.