Deep Learning Assisted Anomaly Detection Support System of Solar Power Plant with Infrared Imagery
Md. Ashif Mahbub, Muhammad Abu Hanif Emon, Md. Hasib Mahbub, Nimur Rahman, ASM Shihavuddin, Sakib Abdul Ahad · 2023
Solar power plants are of significant importance for supporting the movement towards sustainable energy sources. The maintenance of these plants is of utmost importance in order to achieve optimal energy output and ensure their long-term viability. The increasing usage of renewable energy sources necessitates the implementation of improved defect detection systems. The approach of this research is to establish ADSS that employs a deep learning model titled ‘Mirror C-Net’ based fault detection system for the analysis of thermal images obtained from infrared cameras within the solar power plant. The assessment of the ADSS that was built in this research is carried out using a comprehensive dataset, taking into account different environmental conditions and types of faults of solar power plants. The success of the system is assessed using performance indicators such as Accuracy, Precision, Recall, ROC_AUC, and F1-Score. The findings demonstrate that the deep learning-assisted ADSS developed in this study achieves a high accuracy rate of 94.92% in detecting various anomalies of solar modules. Moreover, the suggested model exhibits superior performance when compared to conventional deep learning models. This research has importance due to its potential in reducing operational interruptions and enhancing energy efficiency. This work addresses the integration of cutting-edge technology and sustainable practices within the solar power industry, fostering a more efficient and environment friendly energy production paradigm.