The platform's cloud-based vision system can correctly categorize and locate defects found in thermal image aerial photography. We have recently analyzed a solar rooftop system located in Kaohsiung, Taiwan, which found that 1.28% modules had substring open-circuit conditions after thermal image aerial inspection in 2024, and sampled some of the anomalous modules in 2025 and sent them to the solar laboratory of the Consortium Telecom Technology Center in Luzhu, Kaohsiung, to conduct EL measurements, and the six samples sent for testing were all found to have anomalies, with an accuracy rate of 100%.

EL measurement results initially concluded that the temperature difference between day and night of the solar module of the rooftop system is larger than that of the general ground-type and water-type systems, and that after eight years of operation and thousands of hot and cold cycles, the connection between the battery and the busbar in the module may loosen, resulting in high resistance, and the photocurrent will be biased to cause the diode to turn on due to the high resistance value during the operation. If one diode is turned on, it will cause 1/3 of the module power loss, and so on, and in serious cases, the whole module may not be able to generate power.

This case illustrates the advantages of laboratory analysis after thermal image blanking. The ratio of anomalies can be easily obtained after thermal image blanking (1.28% in this case), and sampling laboratory measurements can be used to obtain an unbiased analysis (which indicates that the anomalies may be due to module reliability issues), and this information can help in the subsequent handling of the problem.