Intelligent meter monitoring in modern utility systems demands robust, scalable, trustable AI solutions. This paper introduces a hybrid AI framework designed for enhanced accuracy, interpretability, and operational trust in meter anomaly detection. The system is designed to meet both data-driven performance and human-centric reliability by deploying specialised agents: (1) A Query Agent for real-time data acquisition; (2) A Text-to-SQL Agent that translates natural language or structured prompts into executable SQL queries for dynamic data retrieval from utility databases; (3) A Rule-Based Enhancement Agent that integrates domain-specific heuristics to refine or validate machine learning outputs; and (4) An Explainability Agent, powered by large language models (LLM), to generate clear, actionable fault narratives reports. By integrating distributed reasoning, hybrid decision logic, and explainable outputs, this system bridges the gap between data-driven automation and the operational demands of modern critical infrastructure—delivering accurate, interpretable, and actionable fault insights for utility operators.
Paper
Full text
Trustable AI Framework for Intelligent Meter Monitoring: A Multi-Agent, LLM-Driven Approach
Semantic Scholar · 2025
Abstract
Intelligent meter monitoring in modern utility systems demands robust, scalable, trustable AI solutions. This paper introduces a hybrid AI framework designed for enhanced accuracy, interpretability, and operational trust in meter anomaly detection. The system is designed to meet both data-driven performance and human-centric reliability by deploying specialised agents: (1) A Query Agent for real-time data acquisition; (2) A Text-to-SQL Agent that translates natural language or structured prompts into executable SQL queries for dynamic data retrieval from utility databases; (3) A Rule-Based Enhancement Agent that integrates domain-specific heuristics to refine or validate machine learning outputs; and (4) An Explainability Agent, powered by large language models (LLM), to generate clear, actionable fault narratives reports. By integrating distributed reasoning, hybrid decision logic, and explainable outputs, this system bridges the gap between data-driven automation and the operational demands of modern critical infrastructure—delivering accurate, interpretable, and actionable fault insights for utility operators.