Evaluation of a Retrieval-Augmented AI Agent for Power Grid Switching Operation Ticket Generation, Safety Screening, and Error Reduction
Abstract
Switching operation ticket drafting and safety verification in power grid dispatching have traditionally relied on manual expertise, creating a persistent gap between drafting quality and review reliability that existing rule-based or single-role AI systems have not fully addressed. A retrieval-augmented, neuro-symbolic dual-agent framework was developed, in which a Generator Agent and a Verifier Agent, instantiated from a shared large language model, operate within a bounded generate–verify–revise closed loop. The Verifier combines deterministic symbolic checking of topology and five-prevention logic with semantic verification against regulatory text. Using real operating-mode change tasks and topology data from the Lishui power grid, this system was compared against a human baseline, a base large language model, and a retrieval-augmented single-pass condition, with eight professionals stratified by experience level. The full system achieved the highest ticket completeness rate (97.8%), the strongest error-screening discrimination (AUC = 0.978), and an 80.2% reduction in observed error rate relative to manual drafting, with 93.3% of drafts converging within two revision cycles. The benefit was found to be larger among less experienced operators. Coupling generation with closed-loop, dual-layer safety verification offers a workable and auditable path toward reducing human error in switching operation ticketing, functioning as a human-authorized support tool rather than an…
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.