AI and energy
Aerial images can reveal whether solar cells are defective, while AI can tell us why. This will ensure faster repair of solar cells—and a greater harvest of green power.
Similar technological solutions are hard to find today. The inspections carried out today are done manually without technology and rarely take place.
“It is not unusual for entire modules to be left standing for long periods without producing electricity. Large solar cell installations are inspected once every 2-4 years — because it is an expensive and time-consuming process,” says Gisele Benatto.
She wants to use drone technology to make it attractive for plant owners to have their solar cells checked more often, so that low power production in solar cells can be detected more quickly.
If this is the case, the solution's AI platform can use the drone images to determine the cause of the low power production, the extent of the power loss, and what needs to be done to solve the problem.
“AI responds immediately with the action needed to maximize the yield of green power,” says Gisele Benatto.
How much more green power can be harvested from a solar power plant if the drone solution is implemented depends on how old the plant is, what condition it is in, and how many inspections are carried out.
However, Gisele Benattos team estimate that a well-functioning system that is inspected once a year could generate 1-2% more energy annually, while an older system in poorer condition could provide between 2-5% more energy.
“If we use our 100 MW test facility in Roskilde as an example, a 1% improvement in annual energy yield would mean that we could supply 230-250 Danish households with electricity for a whole year,” says Mahmoud Dhimish, associate professor and AI expert at DTU Electro.
He adds that another positive effect of the solution is that it enables solar system owners to prioritize repairing or replacing the modules in the system that consumes the most energy first.
At present, the researchers' drone can take pictures of up to 4 modules per minute. In the long term, this number will increase, and the drone will be able to perform the task without human control.
“The plan is to get to the point where the drone can stop at each module, take pictures for a certain number of seconds, and then fly on to the next module,” says Gisele Benatto.
Today, people must stand on the ground and control the drone, just as the researchers must manually activate the electricity in the solar park. Gisele Benatto is researching the development of imaging technology based on light through the phenomenon of photoluminescence. This will enable her to illuminate the solar cells without using electricity, which means that hardware such as cables and wires can be completely avoided.
“It's a solution that eliminates the need for a heavy setup and allows us to obtain high-quality images more quickly,” she says.
However, this research is still at an early stage.
The upcoming plan for the electroluminescence imaging technology is to conduct test inspections of solar cell facilities in the small town Toreby on Falster.
Green energy
Gisele Alves dos Reis Benatto Senior Researcher Department of Electrical and Photonics Engineering garb@dtu.dk
Mahmoud Dhimish Department of Electrical and Photonics Engineering mahdh@dtu.dk