Research on digital twin diagnosis model for the thermal-electric field of high-voltage switchgears.

High-voltage switchgear is a critical component in modern power systems, yet it remains vulnerable to insulation degradation and other faults under complex operating conditions. To address these challenges, a digital twin-based online fault diagnosis method is proposed for high-voltage switchgear, integrating thermal and electric field analysis. A three-dimensional model of the KYN28-12(Z) switchgear is first established, incorporating multi-physics simulations to identify key monitoring regions. Building on this, a digital twin surrogate and information model are developed to enable real-time reconstruction and online characterization of coupled thermal-electric fields. For fault feature extraction, optimized classification tree (OCT) and random forest algorithms are employed, while an enhanced adaptive neural-fuzzy inference system (ANFIS) is constructed for intelligent fault diagnosis. Ultimately, the diagnosis model is trained using a combination of finite element simulation data, experimental acquisition data, and on-site operational historical data, ensuring comprehensive learning of switchgear behaviors under various conditions. And the diagnosis relies on data from the digital twin model to achieve accurate virtual-real mapping of switchgear states, providing theoretical support for intelligent operation and maintenance. Experimental results demonstrate a fault recognition rate of 93.4%, with only a 2.3% accuracy drop under 30% noise, verifying the robustness and reliability of the proposed method.