Fault Location in Active Distribution Networks Using Multiple Measurement-Based Bayesian Learning

Efficient and accurate fault location techniques are beneficial to pinpoint fault position and reduce power outages. Faced with this issue, this paper proposes a novel fault location technique for active distribution networks that utilizes multiple measurement-based Bayesian learning. Specifically, fault location problem in this paper is firstly transformed into solving a block-sparse signal recovery model. In order to enhance robustness in noisy conditions and achieve satisfactory recovery performance, this paper extends the recovery model to a multiple measurement-based model and adopts a block-sparse Bayesian learning (BSBL) algorithm for block-sparse signal recovery. The proposed method requires only a limited number of distribution-level synchronized measurements to be placed, instead of yielding a full observable network. The effectiveness of the proposed method under different noise levels is verified by using IEEE 33-node active distribution system.