20MW Panels Photographed, Analyzed in 2 Days by Using Drone (2)

2MW of panels analyzed in only 3 minutes using AI

2019/05/27 18:14
Shinichi Kato, Nikkei BP Intelligence Group, CleanTech Labo
Print Page

Continued from 20MW Panels Photographed, Analyzed in 2 Days by Using Drone (1)

Thermal distribution images are shot as moving images and the situation is input on the layout drawing on which the flying route is also shown while watching the moving images in the flying sequence. Therefore, almost no time is required to identify locations, which is a big issue for other companies. During manual operation, playing the video of about 2MW of panels and entering the situation are completed at about the same time in approximately 15 minutes. The efficiency was higher compared with services offered by other companies, which normally take about one week.

The company later developed a method to improve the efficiency of the manual operation process by utilizing artificial intelligence (AI) for actual use. The time required to perform the operation for an approximately 2MW plant is reduced to about three minutes by incorporating the method, according to the company.

How many defective panels were discovered by the inspection? The company stated that the number was very small and the panels are in good condition.

The "defect situations" that the company inputs in the solar panel layout drawing are classified into four categories: "hot spot," "abnormality in cluster," "abnormality of entire panel" and "abnormality in junction box."

A "hot spot" means a partially overheated portion. They are normally caused by overheating due to partial failure inside a solar battery cell (power generation element), partial staining of panel surface by bird droppings, etc., or shading by weeds. Hot spots were discovered in 187 locations (Fig. 3).

Fig. 3: Examples of "hot spots" discovered by inspection (source: Energy Solutions)

An "abnormality in cluster" means a defect in a cluster where solar battery cells are connected in series (Fig. 4) and is caused by a defect in internal wiring (inter connector), bypass diode short circuit or other factors.

Fig. 4: Examples of "abnormality in cluster" discovered by inspection (source: Energy Solutions)