Answers are grounded in this paper and cite their sources.
Computational modeling demonstrates accurate hotspot temperature prediction in dry-type transformers using optimized machine learning, highlighting opportunities for digital-twin monitoring.
Transformers are the primary energy-consuming devices in substations. Accurate loss calculation and hotspot temperature prediction are vital for grid stability. This study focuses on a 5250 kVA dual-winding dry-type transformer. We established a magnetic-thermal-fluid multi-physics coupling model. This model considers non-uniform current density in low-voltage windings caused by magnetic leakage. Through field synergy analysis, we explained the relationship between fluid vortices and local hotspots. Furthermore, we proposed a prediction model using the Grey Wolf Optimizer and Kernel Extreme Learning Machine (GWO-KELM). This model uses load rate, ambient temperature, and fan position as inputs. A digital-twin-oriented prototype platform was also developed to support online monitoring. Temperature rise tests show that the simulation error is within 4%. Compared with the backpropagation neural network (BP) and the standard KELM model, the proposed GWO-KELM model tracks hot-spot temperature variations more accurately and achieves lower prediction errors. This research provides a reliable basis for the status monitoring and predictive maintenance of dry-type transformers.
No takes yet. Share an insight, caveat, or question.
Jing et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: