Amber Hickman |
European energy provider E.ON has started using drones and the artificial intelligence (AI) in Microsoft Azure to conduct virtual inspections of power lines in Germany.
The energy supplier’s power grid in the country measures 700,000 kilometres and needs consistent maintenance in order to ensure it can reliably serve homes and businesses. Typically, technicians carry out inspections every five years, checking for defects or signs of wear, and sometimes have to climb the poles. Helicopters are also used once a year to gain a view of entire sections of the grid.
Three of the E.ON distribution system operators in Germany – Mitnetz Strom, E.DIS and Westnetz – decided to improve the process by using aerial drone footage and virtual analysis with the help of AI on Microsoft Azure.
“It was very hands-on work: looking for defects with the naked eye, subjectively assessing them, noting them down, and then entering them into our system,” said Jens Hache, head of the drone project at Mitnetz Strom. “Only after that could we make a decision about whether repairs are necessary or possible. It was difficult to identify defect patterns on the power lines in such a way as to make reliable predictions.”
Technicians now use the drones to take photographs of the power lines in a set pattern. AI technologies is the used to analyse, sort and evaluate them. The images are then loaded into SharePoint, before being moved to an Azure Data Lake.
Next, the images are imported into Grid Vision, the cloud-based software-as-a-service solution developed in Azure by eSmart Systems, a partner of E.ON.
“AI determines whether the distribution system operator images that E.ON sends us show, for example, signs of wear on individual components or major damage to the poles,” said Erik Åsberg, chief technology officer of eSmart Systems. “An expert user then verifies the AI findings before the output is exported back to the E.ON cloud, where it is converted into a checklist.”
Grid Vision is also used by utility providers across Europe and North America, meaning the more images that are analysed by the system, the more the AI learns, creating a worldwide feedback loop that improves the system.
In the future, Mitnetz, E.DIS, and Westnetz aim to improve the efficiency of virtual inspections by enabling drones to fly autonomously across power lines.