Fault Classification for Photovoltaic modules using Thermography and Image Processing

For reliable and efficient operation of solar photovoltaic (PV) system it is necessary to detect and analysis fault. Especially in the case of PV arrays, it is difficult to shut down the modules completely during faults, due to their continuous operation during sunlight and conventional series-parallel PV configurations. This paper observed existing fault detection and classification solutions for PV modules and identifies their challenges and limitations. Since the detection of defects, it is mostly based on the heat radiated from the solar cells, interference of other heat emitting bodies will result in false identification and misinterpretation of the faults. Therefore, it is very essential that the background elements or any external noises need to be eliminated from the image before processing it to fault identification. In this paper, edge detection and Hough transform based image processing techniques were adapted for efficient identification of faults. The processed image is subjected to feature extraction and passed through a classification algorithm for localization and identification of the type of fault. The experiment results depict the training and testing accuracy of the developed technique which are around 94 and 93.1% respectively which are better than conventional methods.