PHOTOVOLTAIC SYSTEM FAULT DETECTION AND CLASSIFICATION BASED ON K-NEAREST NEIGHBOR
Abstract
Faults in photovoltaic systems (PV) have critical damage to the operation
and be the main reason for power loss. To avoid these consequences, the
detection and analysis of PV systems faults must be enhanced effectively and
promptly. In this paper, a fault detection and classification model is developed
and improved. Faults such as open-circuit (OC), line-to-line (L-L), and partial
shading of the systems are detected and classified in K-nearest neighbor (kNN).
Based on the simulation performed in MATLAB/Simulink, the dataset of the
three errors is collected and split at the rate of 75% for the training and 25% for
the testing. The dataset is built with noise conditions as in practice, and the
collecting progress is automated. The absolute error is confined to the range of
0.5% and 5%. It is observed that the result from the proposed model gives high
accuracy of 99.8%.