Andhra Pradesh's Apepdcl Pioneers AI detects incorrect electric poles

Visakhapatnam-based Andhra Pradesh Eastern Coftion Cortion Company Limited (APEPDCL) has introduced an AI-based solution to the long-standing challenge of manually identifying the cause of a power outage. What previously took from a week to several months will be resolved within a day or two, and APEPDCL will make it the first power utility to use AI for this purpose.

Since March 2025, EPDCL's Information Technology Wing has been developing a system that can automatically detect false poles, such as pole tilts, using smartphone cameras. By analyzing the polar images, AI proposes solutions to the authorities, such as repairing and replacing poles, thereby preventing further damage. To develop this system, APEPDCL collaborated with four electrical engineering students at Andhra University and received technical support from Bengaluru-based startup Johnaic. The concept and intellectual property remain in APEPDCL, and the solutions are published without seeking patent protection.

This innovation is expected to increase public safety, reduce power outages and improve reliability. Drone-based monitoring is optional, but APEPDCL has chosen a practical and cost-effective solution that can be used seamlessly by field personnel, including linemen and assistant executive engineers.

“I kept asking myself how new technologies like AI can be incorporated into the daily operations of traditional low-tech Disom. Finally, I found an AI-enabled solution,” AppepDCL CMD I. Hindus Sunday (September 21st).

Previously, technical staff relied on them to identify damaged poles. This is a process in which experienced people require manual inspection of the area, sometimes involving the climbing hills. After collecting photos, each image must be checked individually and may take weeks or months.

“To simplify this complex manual system, we developed an AI-based solution and tested it on 40,000 electric poles. The AI ​​successfully identified the problem with 800 poles with 80% accuracy and then repaired it all. Students at Andorra University played an important role in taking photos and analyzing the data with AI.

“The system is currently integrated into regular operational research, with future expansion planned to detect insulator faults and vegetation growth near the pole. Importantly, the work is now open sourced to encourage adoption by other discoms.