Acoustic Imaging Approach for DC Magnetic Bias Analysis of Power Transformer

DC magnetic bias leads to serious consequences such as increased transformer noise, intensified vibration, incorrect operation of relay protection, and increased transformer losses. Thus, DC magnetic bias analysis is important for power transformer status monitoring. In this paper, a sparse Bayesian learning based acoustic imaging algorithm is proposed for DC magnetic bias analysis by displaying the noise distribution of power transformer. Considering the densely distributed status of the noise sources, a sparse Bayesian learning framework is built to improve the resolution of acoustic imaging. Due to its inherent self-regularization and uncertainty handling properties, the proposed method demonstrates high accuracy performance concerning the false alert ratio, miss detection ratio, and root mean square error. Experiments are carried out to test the estimation accuracy of the proposed acoustic imaging method. The numerical results demonstrate that the proposed acoustic imaging method attains outstanding resolution and accuracy performance, notably in reducing the false alert ratio by more than 4.5%.