A Hybrid Deep Learning Network for Grid-Connected Solar Panel Fault Detection

In recent decades, grid-connected solar power systems have gained enormous interest among industrialists and researchers. Further, the proper installation of these solar power systems and their seamless operation require continuous maintenance to extract efficient and reliable power from these solar panels. In order to achieve efficient power, several power electronics devices as well as numerous control system units, are required. Sometimes, these major elements suffer from faults and operational degradation. Therefore, in this contribution, a fusion of deep learning models has been incorporated, considering a convolutional neural network with a bidirectional long-shortterm memory model to identify and locate the various types of faults in grid-connected solar photovoltaic systems. A comparative analysis is also presented with the available state-of-the-art methods to investigate the efficacy of the proposed technique. Detailed observations indicate the superior performance of the proposed technique for addressing multiclass fault identification and classification. The proposed fault diagnosis scheme could also be applied to other industrial systems for condition monitoring applications.