Heterogeneous Node-Oriented Consistency Optimization Method for Instrument Transformer Monitoring Data in Energy-Storage-Integrated Smart Grids

Abstract

INTRODUCTION:Smart grid digital twins depend on reliable instrument transformer monitoring data for real-time state estimation and downstream operational analysis. However, heterogeneous current
transformers, voltage transformers, capacitive voltage transformers, merging units, and edge terminals often generate asynchronous, biased, and locally inconsistent measurements, which may degrade digital twin reconstruction accuracy and affect downstream grid analytics.

OBJECTIVES: We explicitly distinguish this heterogeneity problem from conventional bad data, random noise, missing values, and topology errors, as it arises from cross-device calibration mismatch and timevarying sensing semantics rather than isolated measurement corruption.

METHODS:To address this problem, this paper proposes a Heterogeneous Topology-Constrained Consistency Optimization (HTCCO) method. HTCCO performs delay-aware temporal alignment, heterogeneity-aware node representation, and time-varying bias correction, followed by trust-weighted aggregation and graph Laplacian-based consistency regularization to enforce topological coherence before state estimation. In this framework, each heterogeneous measurement stream is first mapped into a unified latent space via a learnable compatibility transformation, enabling cross-device alignment prior to aggregation, while trust weights are learned as independent reliability scores rather than purely relative competition. Unlike conventional preprocessing or state estimation methods, HTCCO focuses on front-end measurement correction and consistency enforcement, rather than directly estimating system states.

RESULTS:Experiments on IEEE 14-bus, IEEE 30-bus, IEEE 57-bus, and IEEE 118-bus systems show that HTCCO consistently reduces MAE, RMSE, MAPE, topology consistency residual, and state estimation error compared with WLS-SE, PI-GNN, GAEN, and TAGNN-SE. We further report relative improvements over the strongest baseline across all metrics to explicitly quantify the performance gain. Robustness and ablation results further verify the effectiveness of each module. Finally, we clarify that HTCCO is not directly evaluated in closed-loop storage scheduling tasks; instead, its contribution is a measurement-level consistency layer that improves the reliability of downstream grid optimization and control modules.

CONCLUSION: These results demonstrate that HTCCO provides a reliable front-end data correction and consistency optimization layer for smart grid digital twins, improving the reliability of voltage, current, and power measurements used in downstream monitoring and decision-making tasks.