NELOW-AI Intelligent Leak Management System

By combining a water distribution domain knowledge graph with agentic AI, the system independently understands the hydraulic context of the network and delivers expert-level support—from leak diagnosis to incident response.

Moving beyond conventional monitoring and manual data analysis, the system integrates a water distribution ontology with a large language model (LLM). It structures the physical connectivity of the network and real-time IoT data into a knowledge graph, enabling AI to understand the network as a whole and resolve operational issues like an experienced water network specialist.

Product Features

Autonomous Anomaly Detection & Diagnosis

Continuously monitors real-time pressure and flow data, enabling the AI agent to independently detect and analyze anomalies.

Hybrid Leak Classification

Combines a CNN-based supervised learning model for leak classification with an autoencoder-based unsupervised learning model for anomaly detection.

Hydraulic Correlation Analysis

Uses knowledge graph–based network topology analysis to logically identify suspected leak sections based on changes in water pressure.

Conversational User Interface (LUI)

“Show me where the pressure dropped the most last night.” Responds instantly to natural-language requests with maps, charts, and reports.

Knowledge-Grounded Guidance (RAG)

Generates practical response guidance for field issues based on repair manuals and historical incident records.

Key Benefits

Autonomy Through Self-Reflection

Rather than simply repeating predefined tasks, the system evaluates the relevance of its own outputs and continuously improves its reasoning and operations.

Expert-Level Operational Support

AI tools incorporate the expertise of leak detection and hydraulic analysis specialists, providing operational support that approaches expert-level decision-making.