Fault Location in Active Distribution Networks Using Block Sparse Bayesian Learning

Abstract

Fault location is one of critical issues in the operation of distribution networks. Timely and accurate fault location can avoid time-consuming examinations and huge economic losses. Facing the issue, this paper introduces a block sparse Bayesian learning (BSBL) based fault location approach for active distribution networks. This approach utilizes the synchronized phasor measurement information provided by a limited number of phasor measurement units (PMUs) placed in the distribution network. By formulating the fault location problem into a block sparse signal estimation model, a BSBL framework is adopted to recover the sparse signal and thus determine the fault type along with accurate fault position. The utilization of BSBL method is able to exploit both block structure of the sparse signal and correlation among the non-zero variables, which helps further improve the recovery performance. In addition, the principle of the PMU placement is described in this paper, which offers an insight into the application of fault location using only a few PMUs. The effectiveness of the proposed approach has been verified by testing different fault types, fault resistances, fault positions and noise levels.