Distribution Network Fault Identification and Localization based on MCS-SBL Algorithm

With the continuous development of the power system, the reliability and security of the distribution network are more and more emphasized. In this paper, a new method for fault identification and localization in distribution networks is proposed based on the multi-response sparse Bayesian learning (MCS-SBL) algorithm. By simulating the IEEE33 standard node system, the mathematical model between node voltage variation and fault current is established, and the sparsity constraint is used to reconstruct the fault current. The experimental results show that the proposed method can effectively identify and localize multiple fault types, and the accuracy of fault localization remains above $90 \%$ under different transition resistance conditions. In addition, the MCS-SBL algorithm shows good performance in dealing with multiple faults, and is able to accurately localize the fault region, which significantly improves the efficiency and reliability of fault detection. The results of the study provide strong technical support for distribution network fault management and have a wide range of application prospects.