Digitizing electricity generation in solar panel farms with digital twin and operating with high efficiency

Renewable energy sources, particularly solar energy, have witnessed significant growth in popularity and production capacity. Over 1000 GW of photovoltaic panels have been installed and operational worldwide by the end of 2022. However, ensuring high-energy production efficiency, preventing, and detecting errors in the system have become increasingly urgent. The non-linear nature and unpredictability of solar power production make predictability and instantaneous change in production challenging. Therefore, this thesis aims to enhance the traceability of solar power plants by developing a digital twin and creating a platform utilizing the necessary tools to optimize the system's performance. Initially designed in Matlab, the solar power plant was moved to a simulation environment to evaluate its operating characteristics and refine its design. The data collected were used to train machine learning algorithms, which are employed to monitor the system, predict future scenarios, and notify users when an unusual event or power loss occurs. Additionally, an algorithm was created to address contamination issues and optimize panel cleaning. The integration of these methods and programs, along with flexible interconnection, provides enhanced data visualization, improving traceability. It is hoped that this thesis will contribute to researchers and plant owners who want to examine solar power plant behavior and increase the efficiency and traceability of solar power plants.