APEPDCL Rolls Out AI-Powered Grid System To Predict And Prevent Outages
This project emerged from collaborative efforts between APEPDCL’s IT engineers, students from Andhra University, and a Bengaluru-based private technology firm: Reports
VISAKHAPATNAM: Andhra Pradesh Eastern Power Distribution Company Limited (APEPDCL) has revolutionised power grid maintenance through the strategic deployment of artificial intelligence (AI) technology, transforming how electrical infrastructure failures are prevented and detected.
Rather than investing in costly drone-based monitoring systems, APEPDCL has implemented a practical, budget-conscious solution that integrates seamlessly with existing field operations. This approach enables linemen and assistant executive engineers to leverage advanced technology without disrupting established workflows.
APEPDCL's chairman and managing director Prudhvi Tej Immadi, in a press release, said, “I envisioned a system that could identify damaged power poles before catastrophic failures occur, and this has led to embedding AI technologies into traditional DISCOM operations, which historically lacked high-tech integration.”
This project emerged from collaborative efforts between APEPDCL’s IT engineers, students from Andhra University, and a Bengaluru-based private technology firm.
In March 2025, APEPDCL deployed an AI server within its data centre, creating an automated system for infrastructure fault detection using smartphone cameras. The technology accurately identifies leaning poles and tilted cross-arms, revolutionising routine operational surveys with enhanced precision and efficiency.
Students received training in GPU programming and AI workflows under expert supervision, while the IT team, led by Van Srinivas, coordinated with linemen to capture training data.
The system now functions within regular operational surveys, with planned expansions targeting insulator defect detection and vegetation growth monitoring.
Beyond pole detection capabilities, APEPDCL has identified smart meter data analysis as a high-impact application area. The utility’s IT team is leveraging this data for multiple operations, including faulty meter detection to prevent revenue losses, real-time granular load-analysis graphs to identify low-voltage areas, and improved load prediction accuracy for proactive grid management.
The roadmap also includes deploying a large language model (LLM) in partnership with the private technology firm. This LLM, securely hosted within APEPDCL’s firewall, will enable officers to query systems using natural language, access Standard Operating Procedures, and receive immediate corrective action recommendations.
Current development is focused on local demand forecasting models for feeder and distribution transformer levels, enabling early overload detection and optimising manpower deployment for timely load-shifting operations.
