Integrated Approach for Dust Identification and Deep Learning-Based Classification of Photovoltaic Panels

The accumulation of dust on photovoltaic (PV) panels faces significant challenges to the efficiency and performance of solar energy systems. In this research, we propose an integrated approach that combines image processing techniques and deep learning-based classification for the identification and classification of dust on PV panels. The image processing algorithms are utilized to detect and segment dust particles on the panel surface. Various pre-processing techniques are applied to enhance the accuracy of dust detection, including image enhancement and noise reduction. The algorithm effectively distinguishes between dust particles and the panel surface, enabling precise localization of dust contamination. To achieve accurate classification, a deep learning model based on a convolutional neural network (CNN) architecture is trained. The model is trained on a dataset consisting of labeled images of dust and dust-free panels. The training process optimizes the model's parameters to achieve high classification accuracy. The proposed deep learning architecture attains an accuracy of 98.5%, a sensitivity of 98.5%, a specificity of 97.34%, and a precision of 98.1%.