Design of a remote operation and maintenance assistance system for substations based on augmented reality and three-dimensional modeling

Résumé fourni par la source

Substation operation and maintenance (O&M) increasingly depends on heterogeneous sensing, three-dimensional asset models, remote expert collaboration, and auditable automation support. However, conventional mobile maintenance applications still separate equipment state, spatial context, safety procedures, expert guidance, and instruction confirmation, which weakens the timeliness and traceability of field decisions. This paper designs a remote O&M assistance system that integrates augmented reality (AR), lightweight 3-D modeling, edge-cloud synchronization, risk-aware decision support, and an automatic AR-instruction return-and-acknowledgement loop. A BIM-mesh-NeRF representation and two-stage visual-inertial/point-to-plane registration strategy stabilize spatial overlays. Before a remote instruction is released, the system verifies the work-ticket binding, operator identity, equipment state, model version, and registration confidence; the field acknowledgement, execution evidence, exception state, and timestamps are then returned automatically to the remote side. In a controlled 220-kV demonstration-bay test, the proposed system reduced mean overlay error from 64 to 27 mm, end-to-end latency from 168 to 92 ms, task completion time from 27.4 to 18.6 min, and report preparation time from 16.2 to 5.1 min. Median instruction-confirmation latency decreased from 6.8 to 1.4 s, while the unconfirmed-instruction rate fell from 6.7% to 0.8%. These results indicate that AR can support a closed, traceable maintenance-assistance loop without becoming a direct path to protection, trip, or interlocking control.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.