The platform’s cloud-based vision system can accurately classify and locate defects detected through thermal aerial inspections. Recently, we analyzed a rooftop solar power system in Kaohsiung, Taiwan. After conducting a thermal aerial inspection in 2024, we found that 1.28% of the modules had substring open-circuit issues. In 2025, several abnormal modules were sampled and sent to the Solar Laboratory of the Telecommunication Technology Center in Luzhu, Kaohsiung for EL (electroluminescence) testing. All six tested samples exhibited abnormalities, achieving a 100% detection accuracy.

Preliminary EL measurement results indicate that rooftop solar modules are subject to larger day–night temperature differences compared to typical ground-mounted or floating systems. After eight years of operation and thousands of thermal cycles, it is likely that the connections between the solar cells and the busbars inside the modules have loosened. This loosening increases the resistance, and during operation, the elevated resistance causes a forward bias on the bypass diode, leading it to turn on. If even one bypass diode is activated, the module will lose one-third of its power output. In more severe cases, the entire module may stop generating power.

This case illustrates the advantages of combining thermal aerial inspection with laboratory analysis. Thermal drone inspections can quickly and easily determine the abnormality rate (1.28% in this case), while laboratory testing of sampled modules provides objective and impartial analysis (in this case, indicating that the issue may stem from module reliability). These insights are valuable for guiding subsequent troubleshooting and corrective actions.