Classification of causes of broken solar panels in solar power plant

In this paper, we report various methods for classifying faults that use the data of string measurement devices used for continuously monitoring solar power panels remotely. Low power generation of solar panels is caused not only by panels being broken but also by shadows cast by structures, weeds, etc. If these failures can be classified by using the data of remote string measurement devices, it is expected that the number of unnecessary repairs will be reduced, making preparations for possible failures more efficient. We focused on low-open circuit voltage cluster failure, shadows, and weeds, which often decrease power generation at solar panels, and we examined these classification methods with string measurement data. Furthermore, a failure classification flow was created by combining various failure detection methods. When comparing this flow with the results of drone inspection, the accuracy rate was 74.0%.