Fault Line Selection Strategy for Distribution Networks Based on Dynamic and Accurate Measurement of Zero-Sequence Current—A Data–Model Hybrid-Driven Method
Abstract
The measurement accuracy issue of zero-sequence current transformers (CTs) has long been a critical factor restricting the accuracy of fault line selection in distribution networks. Although existing research methods are relatively mature in theory, their on-site application is limited by the measurement precision of zero-sequence CTs. To address this problem, this paper proposes a fault line selection strategy for distribution networks based on dynamic and accurate measurement of zero-sequence current. Firstly, from the data perspective, this paper analyzes the fault characteristics of various electrical quantities in different operation stages of distribution networks. Combined with system characteristics, an accurate measurement method for zero-sequence current amplitude is subsequently put forward. Afterwards, a distribution network fault line selection algorithm optimized by an attention mechanism-based multi-scale convolutional neural network is constructed. Finally, verification results based on the IEEE standard test system demonstrate that the proposed method enhances the capabilities of feature extraction and side information aggregation, realizes efficient and accurate localization of faulty lines, and exhibits strong robustness under noisy conditions.