Detection and Identification of faults in a PV Module using CNN based Algorithm

• Published in: 2022 3rd International Conference for Emerging Technology (INCET) pp. 1 - 5
• Main Authors: Prajapati, Nikhil, Aiyar, Ramanansri, Raj, Ayush, Paraye, Milind
• Format: Conference Proceeding
• Language: English
• Published: IEEE 27.05.2022
• Subjects: Bypass diode; CNN; Fault detection; Fault diagnosis; Faults; Heating systems; Hotspot; Image processing; Photovoltaic module; Photovoltaic systems; Production; Real-time systems; Thermal Image
• DOI: 10.1109/INCET54531.2022.9825452

Additional power losses, hotspots, and varying irradiances across PV modules can all be caused by faults in PV arrays. As a result, there is a loss of production and a decrease in generation efficiency. If the faults are not addressed, they may spread to neighbouring modules, resulting in the full failure of PV strings. Hence to overcome these faults and its repercussions, it is necessary to have a system in place that would help identify the faults efficiently and accurately. The thermal images dataset contains a set of images shot through a drone and it is detected using the box detection algorithm. The detection of the faulty part of the panel and classification of four different types of faults viz. temporary hotspot fault, permanent hotspot fault, bypass diode fault and crack/wear and tear, are done using the CNN learning algorithm which is YOLO (You Only Look Once).