| Work Detail |
Researchers in Saudi Arabia have compared the performance of land-based photovoltaic plants with that of offshore solar installations and found that floating installations benefit from the cooling effect of seawater. Researchers at King Fahd University of Petroleum & Minerals in Saudi Arabia have conducted a comparative study of experimental floating solar photovoltaic (SFPV) systems and ground-mounted solar photovoltaic (GSPV) systems using an artificial intelligence (AI) system that predicts the surface temperature and power output of both configurations. “Our findings contribute to the advancement of AI techniques to improve smart grid integration and operational efficiency of floating solar installations, aligning with the innovation goals of sustainable energy development in Saudi Vision 2030,” corresponding author Dr. Mohamed E. Zayed told pv magazine . “SFPV and GSPV systems are installed and tested under the same climatic conditions in Azizia, Kingdom of Saudi Arabia, and thoroughly evaluated with respect to the electrical power produced, PV surface panel temperature, PV-DC voltage, PV-DC current, and energy performance and efficiency,” the team explained. “The second objective of this study focuses on the application of advanced artificial intelligence models to predict electrical power generation and PV surface panel temperature in both SFPV and GSPV systems, an area that has not been rarely investigated.” Both the SFPV and GSPV systems consisted of two bifacial panels with a maximum power of 545 W. Both systems also included an inverter, a battery, and a set of data loggers and measurement devices. The SFPV system was installed 25 m off the coast of the Gulf of Bahrain, in Azizia, at a depth of 1.5 m, while the GSPV was installed nearby, on land. The SFPV also used a wooden frame, reusable plastic drums, a stainless steel support structure, steel cables, hooks, and concrete anchor blocks. The analysis showed that the long-term average ambient temperature fluctuated between 15.35°C in January and 36.0°C in July, with a relative humidity that reached 31.65% in June and peaked at 68.23% in December. The daily global horizontal solar intensity ranged from 3.30 kWh/m2/day to 7.74 kWh/m2/day, with a global average of 5.64 kWh/m2/day throughout the year. Furthermore, the average wind speed at 10 m above sea level ranged from 3.71 m/s in October to 5.42 m/s in June. Measurements of both devices were conducted in June 2024 and showed that the SFPV system improved the average photovoltaic electrical power and cumulative daily net electrical energy by 59.25% and 69.70%, respectively, compared to the ground-mounted system. This was due, in part, to the cooling effect of seawater. While the average temperature measured on the surface of the GSPV was 58.40°C, the SFPV had a temperature of 39.5°C, a reduction of 32.36%. To predict the capabilities of these systems, the group combined the Brown Bear Optimization Algorithm (BBOA) with the Long Short-Term Memory (LSTM) technique. BBOA is inspired by the natural behaviors of brown bears and is used to fine-tune the LSTM models hyperparameters. Hyperparameters are the external settings established before the LSTM begins its learning process, which govern its operation. LTSM then uses its pattern-understanding capabilities to predict outcomes. The dataset was partitioned using a 70/30 split, with 70% of the dataset allocated to training and 30% to testing, the group explains. The models input variables include characteristics such as time, solar radiation, photovoltaic current, photovoltaic voltage, and ambient temperature, while the target outputs are electrical power and photovoltaic surface temperature. The LSTM-BBOA was then compared with three other models: Light gradient-boosting machine (LightGBM), LSTM-only, and gated recurrent unit (GRU). The results showed that the LSTM-BBOA model achieved superior robustness on both SFPV and GSPV systems. For SFPV power, it achieved a deterministic coefficient (R²) of 0.9998. For compression, the LSTM-only model achieved 0.9966, and the LightGBM, 0.9844. The analysis showed that the hybrid LSTM-BBOA exhibited robust performance, with minimum mean absolute error (MAE), root mean square error (RMSE), and coefficient of variation (COV) values ??of 0.4884, 0.5031, and 0.1938 for SFPV power output predictions. The standalone LightGBM, meanwhile, showed maximum MAE, RMSE, and COV values ??of 5.7036, 12.6872, and 20.3577, respectively. “The LSTM-BBOA model achieved maximum efficiency coefficient (EC) and overall index (OI) values ??of 0.9998 and 0.9931, respectively, surpassing the LSTM model’s scores of 0.9969 and 0.9472 for SFPV power production,” the scientists concluded. “In comparison, LightGBM recorded the lowest EC and OI values, at 0.9844 and 0.9190, respectively.” Their findings were presented in “ Benchmarking reinforcement learning and prototyping development of floating solar power system: Experimental study and LSTM modeling combined with brown-bear optimization algorithm ,” published in Energy Conversion and Management . |
