Chinese scientists have developed and demonstrated a silicon solar panel defect detection system capable of operating in all weather conditions. The uniqueness of the project is that the existing methods for assessing the quality of panels cannot be applied in daylight conditions.
About 90% of all solar panels in the world are manufactured from silicon, but they often have defects that occur during production, installation or maintenance. These defects significantly reduce the already not very high efficiency of the panels, so their quick and easy detection is an important task.
The solution was proposed by scientists from the Nanjing University of Science and Technology (PRC) – they published the results of their project in the scientific journal Applied Optics. The system, combining the latest hardware and software solutions, makes it possible to clearly display solar panel defects, even in bright light, and to analyze them.
During operation, the system passes a modulated electric current through the panel, causing it to emit light, which turns on and off at a high frequency. Then, using a high-frequency InGaAs sensor (based on a compound of indium, gallium and arsenic), a sequence of images is taken. To compensate for the noise generated by sunlight, a filter is used to limit detection to wavelengths around 1150 nm (infrared).
“The very high frame rate allows a large number of images to be collected so that more changes in the images can be discerned. The development is based on a new algorithm that distinguishes between modulated and unmodulated elements in a sequence of images, and then increases the difference. This makes it possible to clearly detect solar panel defects in high light conditions. “, – explained the work of the system Sheng Wu, one of the authors of the project.
To test the system, scientists connected it to solar panels based on monocrystalline and polycrystalline silicon. The test showed that the system successfully detects defects on silicon solar panels under conditions of radiation intensity ranging from 0 to 1300 W / m², which corresponds to illumination from complete darkness to bright sunlight.
Currently, the authors of the project are working on improving the software, thanks to which it will be possible to reduce the impact of noise in order to obtain more accurate images. In the future, they plan to use artificial intelligence algorithms to automatically identify types of defects and further simplify the verification process.