VISAKHAPATNAM: The Andhra Pradesh Eastern Power Distribution Company Limited (APEPDCL) has developed an Artificial Intelligence (AI)-based system to detect faulty power poles, aiming to improve public safety and reduce outages. The initiative was carried out in collaboration with students of Andhra University and Bengaluru-based startup JOHNAIC. The AI model, deployed on an on-premises server at APEPDCL’s data centre in March 2025, identifies leaning poles and tilted cross-arms using images captured on smartphone cameras.
Linemen were trained to collect field data, which was later used to build the system with guidance from JOHNAIC CEO Sasank Chilamkurthy. The project was led by APEPDCL IT engineers and overseen by CMD Prudhvi Tej Immadi. Explaining the approach, Prudhvi Tej said, “We wanted to embed AI into everyday DISCOM operations. The model proves that home-grown innovation can support our manpower and strengthen grid reliability.”
The system has been integrated into operational surveys and is being expanded to detect other issues such as insulator defects and vegetation growth. The project has also been open-sourced to enable adoption by other DISCOMs. “Alongside pole detection, APEPDCL has been using smart meter data for operational improvements.
The IT team has developed methods to detect faulty meters, prevent revenue losses, generate real-time load-analysis graphs, and identify low-voltage pockets. The data supports accurate load forecasting at feeder and distribution transformer levels, enabling timely intervention to avoid overloads,” Immadi elaborated.
As part of its roadmap, APEPDCL is working with JOHNAIC and Fluent Grid to deploy a large language model (LLM), hosted securely within its firewall.
The system will allow officers to access operating procedures, query systems in natural language, and receive corrective recommendations. Data from smart meters, feeders, and distribution transformers is being consolidated into a central DataLake for advanced AI applications.
Chilamkurthy said, “With JOHNAIC 140 and the Data Lake, we will provide APEPDCL with a private LLM that predicts failures, guides operators, and ensures continuity of power supply within the utility’s own infrastructure.”