Data-driven hierarchical adaptive protection scheme for recloser coordination in active distribution networks

Recent advances in data-driven protection have created new opportunities for adaptive protection of active distribution networks (ADNs). However, for complex multi-feeder systems, training a single global model often results in high computational complexity, overlapping fault signatures among feeders, model confusion, and reduced prediction accuracy, particularly under varying fault resistance conditions. Therefore, this paper proposes a self-adaptive protection strategy for automatic recloser setting that overcomes these limitations through a hierarchical framework. Fault transmitters installed at feeder branches provide rapid hardware-based fault pre-localization, enabling the protection algorithm to identify the faulted feeder before any decision process is executed. Consequently, only a lightweight localized data-driven linear regression model trained with data from the corresponding protection zone is activated, significantly reducing the search space while improving estimation accuracy and computational efficiency. The localized model estimates the prospective fuse current under different fault conditions, including varying fault resistance levels, to adaptively adjust the recloser operating time according to the fuse clearing time. The proposed method is validated on the IEEE 33-bus testbed under various fault locations, distributed generation (DG) penetration levels, and fault resistances. Simulation results demonstrate accurate fuse current estimation under diverse operating conditions. In addition, the hierarchical protection framework significantly reduces the operating time of the recloser for permanent faults on both the main feeder and downstream feeders, where the recloser operates as the primary and backup protection device, respectively, while preserving the fuse-saving philosophy and reliable fuse–recloser coordination. Overall, the proposed framework provides an effective solution for adaptive protection of modern ADNs.