May 24, 2022


Uttejh Reddy

From “How?” to “Wow!”

The revolutionary impact of AI in the Powerline Industry

Electricity is an essential part of modern life and the foundation of a prosperous economy. Power outages are rare in affluent countries because of the high quality of electric asset maintenance. This is due to the fact that the electric utilities engage in regular inspection of the assets in order to ensure an uninterrupted power supply. If you have not given powerline inspection much thought before, you will probably wonder how power lines are inspected today. 

A generation of no change

For over half a century, utilities have employed helicopters with senior inspectors onboard or employees who manually walk the grid to report the issues. This suboptimal and risky process has continued for decades, however, there is a shift in momentum due to the advent of new technologies such as drones and sensors. This has sparked a surge in utility companies attempting to digitalize their workflow for asset management. 

The challenge, however, is the sheer volume of data (images of power lines and its surroundings) that must be manually reviewed in order to find defects. This has lead to delayed adoption of the technology, but also to a massive opportunity for the introduction of Artificial Intelligence (AI) - that can help automate and optimize power line inspections by using new computer vision technology to scan images in the search of defects. 

The dominance of AI

In the previous decade, many companies in other industries have changed their focus to AI. This transformation has been attributed to the tremendous increase in computational capacity, but it is further fuelled by the development of Deep Neural Networks.

Skyqraft has collaborated with major utility companies to demonstrate the value of Artificial Intelligence. To give you a concrete example of a use case, Skyqraft’s AI finds damaged insulators on images taken by drones.

In this specific case, thousands of drone images were uploaded to Skyqraft’s software to analyze, report, and visualize defects on a 60 km stretch of the grid. Within minutes, the software identified the first critical defect, a cracked ceramic insulator.

In total, the software identified 3 critical insulator problems on the stretch in just under 28 minutes. The time it takes to upload the images to the server takes far longer than the time needed by the AI to process the images. Once all the defects were flagged by the AI, an experienced human image inspector went through the flagged images and verified that they were indeed critical issues that needed to be fixed. This step is important. The AI does not decide what needs to be repaired, instead it just gives suggestions and a human can then decide calmly what defects that should be prioritized. 

Finally, from the Skyqraft’s software, a list of all the defects were seamlessly exported to the client's Asset Management System. The work orders are then done in that system. Getting a work order with image support and GPS location is extra valuable for bringing the correct tools, correct repair material and drive to the correct location. 

For reference, it takes human image inspectors approximately 210 hours to complete the same stretch that took the AI just 28 minutes. Needless to say, the potential of AI and the value to clients is enormous. AI can find more defects, report them with a higher accuracy than human image inspectors and also do it at a lower total cost. 

If you interested in learning more about Skyqraft, don’t hesitate to reach out at
Uttejh Reddy
Lead Data Scientist