Yu Li, Dingwen Hu, and Rui Yang
View Full Paper
Photovoltaic array, fault diagnosis, GAN, CNN
This paper briefly introduces the generative adversarial network (GAN) algorithm for enhancing fault training samples and the con- volutional neural network (CNN) algorithm for diagnosing faults in photovoltaic arrays. Simulation experiments were conducted. Photo- voltaic array equipment was used to collect actual photovoltaic array fault data. Then, the GAN algorithm was used to enhance the sam- ple data, and the performance of the GAN algorithm was indirectly evaluated using the CNN algorithm. Finally, it was compared with the support vector machine and back-propagation neural network (BPNN) algorithms. The results showed that the GAN algorithm could effectively enhance small sample data and the CNN algorithm could diagnose faults in photovoltaic arrays more accurately. Fea- ture indicators such as output current, output voltage, load voltage, and load current were important for diagnosing the types of faults in photovoltaic arrays. The novelty of this paper lies in using the GAN algorithm to enhance the fault sample data, so as to solve the problem of the lack of high-quality photovoltaic fault data in reality, and using the CNN algorithm to identify photovoltaic faults.
Important Links:
Go Back
