© 2016 IEEE. Current-voltage (I-V) curve traces of photovoltaic (PV) systems can provide detailed information for diagnosing fault conditions. The present work implemented an in situ, automatic I-V curve tracer system coupled with Support Vector Machine and a Gaussian Process algorithms to classify and estimate abnormal and normal PV performance. The approach successfully identified normal and fault conditions. In addition, the Gaussian Process regression algorithm was used to estimate ideal I-V curves based on a given irradiance and temperature condition. The estimation results were then used to calculate the lost power due to the fault condition.
Jones, C. B., Martinez-Ramon, M., Smith, R., Carmignani, C. K., Lavrova, O., Robinson, C., & Stein, J. S. (2018). Automatic fault classification of photovoltaic strings based on an in situ IV characterization system and a Gaussian process algorithm. In 2017 IEEE 44th Photovoltaic Specialist Conference, PVSC 2017 (pp. 1–6). Institute of Electrical and Electronics Engineers Inc. https://doi.org/10.1109/PVSC.2017.8366372