15 January 2023 in Electric Power Systems Research
10 December 2025 in Sustainable Energy, Grids and Networks
18 September 2026 in Sustainable Energy, Grids and Networks
8 June 2026 in IEEE Transactions on Instrumentation and Measurement
To address the challenges of large measurement errors, extensive data missing, and weak observability in distribution networks, this paper proposes a hybrid data-driven and model-driven topology identification method. This method integrates multi-period partial switch measurements along with their remote signaling data and load measurements to construct a Dual-tower Transformer (DTT) imputation-identification model. The left tower leverages the self-attention mechanism to fuse temporal measurements and missing masks for estimating missing values, while the right tower employs the multi-head attention mechanism to integrate imputed estimates with remote signaling data for topology identification. The two towers are jointly optimized through a composite loss function incorporating normally-open switch constraints, and through multiple iterative updates, accurate topology identification is achieved while maintaining imputation accuracy. To enhance the reliability of identification results, a correction strategy for suspicious switch states is proposed. The sigmoid output values of the DTT model are used to calculate switch prediction confidence, low-confidence switches are filtered based on a threshold, and efficient correction is realized by combining state estimation with an enumeration method. Finally, the effectiveness of the proposed method is validated through IEEE standard test cases and practical examples.