Answers are grounded in this paper and cite their sources.
Temperature distribution inside gas-insulated transmission lines (GILs) is a key indicator of operational status and potential faults. However, the high electric field and pressurised environment inside GILs make internal sensor installation impractical. Traditional simulation methods are too computationally intensive for real-time monitoring and lack the capability to reconstruct internal temperature fields from external measurements. To address these challenges, a multi-model collaborative digital twin (DT) approach is proposed, integrating physical mechanisms, data-driven modelling, and geometric representation for real-time thermal field prediction. A high-fidelity multiphysics simulation model ensures physical consistency and serves as the source of interpretable training data. A high-accuracy data-driven surrogate model—the Delta-Enhanced Cross-Attention Parallel Transformer-LSTM Network (DE-CAPT-LSTM)—is developed to enable rapid inversion from sparse external sensor measurements to the full-field internal temperature distribution. A 3D geometric model, combined with a visualisation engine, supports dynamic rendering of thermal conditions. The resulting DT system achieves high computational efficiency by leveraging the physical interpretability of the simulation model and the rapid inference capability of the neural network surrogate. With only limited external sensor data, it enables comprehensive and real-time monitoring of internal thermal behaviour, providing effective support for fault diagnosis and preventive maintenance of GILs.
No takes yet. Share an insight, caveat, or question.
Luo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: