L R AS Published on Sunday 23 January 2022 - n° 390 - Categories:panels
Drones and artificial intelligence analysis to detect signs
Solar power plant managers use artificial intelligence (AI) to analyse the operation of solar panels and identify any faults. This is a good alternative to sending out an inspection team,
which is slow, costly and not very accurate.
AI-assisted inspection can be carried out by a drone or an unmanned aerial vehicle (UAV). It performs quality control of panels using aerial imagery.
The images can be processed either in the cloud or in the aircraft. The algorithms will tell the controller which photovoltaic panels show visible signs of defects.
By classifying them automatically, inspectors can analyse the installation in a matter of hours rather than days of on-site visits. Locating faulty panels improves efficiency
Deep learning
The most common type of algorithm used to inspect solar panels is a deep learning algorithm. It uses a neural network to learn how to solve a task. It is made up of interconnected layers. They learn to recognise defects in solar panels from images.
Building a deep learning network requires a training period. It involves viewing large sets of image data, which are categorised as good or bad. The solar farm operator will mark each image as either defective or not defective, so that the neural network learns to identify the two types of panels.
Once recognition has been learned, the machine can be used to inspect solar panels in images collected from a solar farm. The neural network will identify any defective solar panels in the image and provide a classification (defective or not defective).
This requires: a°) providing the computer with training data, so that it can learn to detect solar panel defects. It needs a large set of labelled image data.
b°) The challenge is to provide all types of panels. Solar farms may install hundreds or even thousands of different types and models of solar panels, each with its own characteristics, such as size, shape and colour. For each plant, data on the panels that make up the plant must be provided.
c°) The algorithms developed to detect faults in solar panels are not 100% accurate. This means that a small number of solar panels may be incorrectly classified as faulty.
Overall, AI is a very powerful tool for solar farm operators. It should be integrated into their maintenance routine.
https://www.pv-magazine.com/2022/01/21/how-artificial-intelligence-can-be-used-to-identify-solar-panel-defects-2/
PV Magazine of 21 January 2022
