Arc Fault Diagnosis of DC Microgrid Based on t-SNE Feature Fusion

As a novel energy system, DC microgrids have garnered global attention. To address the challenge of detecting series arc faults in DC microgrids, this paper proposes an approach that integrates variational mode decomposition (VMD) with t-distributed stochastic neighbor embedding (t-SNE) for arc fault identification in DC microgrid systems. By extracting the fundamental time-domain and frequency-domain characteristics of arc fault currents in DC microgrid systems and analyzing the time-frequency domain features using VMD, a high-dimensional hybrid-domain feature vector is constructed. Subsequently, t-SNE is employed to fuse the fault features through dimensionality reduction. Finally, the support vector machine (SVM) is used to complete the fault diagnosis.