AUTOMATIC FAULT DETECTION OF PHOTOVOLTAIC ARRAY BY CONVOLUTIONAL NEURAL NETWORKS DURING AERIAL INFRARED THERMOGRAPHY

摘要

Aerial Infrared Thermography (IRT) is a non-destructive and cost-effective method for detecting faults in large-scale photovoltaic (PV) power plants. However, the visual assessment of the images captured by aerial IRT, and the analysis of a large number of image frames is very time-consuming. This paper proposes a method for detecting and classifying faults on PV modules, through aerial IRT images, combining Digital Image Processing (DIP) and Convolutional Neural Networks (CNNs) algorithms. With the results obtained so far, the IR images acquired were successfully processed with DIP techniques to detect the faults of PV modules in the power plant that are used as samples for training the CNN. The developed neural network algorithm can detect faults on the aIRT images and classify them in three categories: disconnected substrings, hot spots, and disconnected strings. The results have demonstrated that the method is effective in detecting and classifying faults, and it is an important step for the full automation of aIRT inspection.