Dust Detection on Solar Photovoltaic Panels Used in Optoelectronics with Convolutional Neural Network-Based Deep Learning Models

Abstract

Solar photovoltaic panels, one of the optoelectronic device types, contain a large number of photovoltaic cells. The maintenance of these solar panels with photovoltaic cells is very important for the efficiency of the energy obtained from the panel. As time passes, dust may form on the panels due to various weather conditions and environments where the panels are located. In order to maintain the panels in a timely manner and increase energy efficiency, this study aims to detect the dust on the panels. For this reason, an open source dataset consisting of normal, clean and well-maintained solar photovoltaic panels and solar photovoltaic panels containing dust was used. Since the amount of the dataset is small and the amounts in the classes are unbalanced, firstly, various data augmentation operations were performed to increase the number of data amounts and make it balanced. In order to use this balanced dataset in the classification phase with deep learning models, the dataset was divided into 80% training and 20% testing. After this process, a total of four deep learning models based on convolutional neural networks, including MobileNetv1 for dust detection in solar photovoltaic panels and ResNet models with three different number of layers, were used. During these processes, two different optimization methods were used to train each model. As a result of these detection studies, the highest accuracy value was found to be 0.993 in the ResNet model, which was trained using the AdamW optimization method and had 18 layers.