Intelligent power forecasting of HIT and amorphous PV technologies using ANFIS and neuro-fuzzy RNN models

The operating module temperature, wind flow, and radiation intensity across the PV module have a larger effect on the PV module's seasonal performance. Ultimately, it influences the power output of the module. Using two distinct PV technology modules Amorphous-Silicon (a-Si), and Hetero-junction with Intrinsic Thin-layer (HIT) set up at the National Institute of Solar Energy (NISE), Gurgaon site in India, this work outlines a novel method for forecasting the effects of module temperature, efficiency and power on seasonal performance variation to support sustainable solar energy utilization and efficient energy management. This work develops ANFIS and hybrid RNN models and compares them with empirical formulas which have been already established to improve the accuracy of photovoltaic power temperature and power forecasting and enhance reliability of sustainable energy systems. Field data collected during the winter, summer, and monsoon seasons were used, accounting for such variables as wind speed, irradiance, and ambient temperature to understand seasonal variability and its implications for sustainable solar power generations. For a-Si technology, the Root Mean Square Error (RMSE) values for temperature prediction varied from 1.34 to 3.35 in winter, 2.82–6.59 in summer, and 2.30–9.87 during monsoon conditions. Similarly, for HIT technology, the RMSE values ranged from 2.16 to 3.66 in winter, 2.44–4.61 in summer, and 1.75–9.41 in monsoon season. For power forecasting, the Hybrid RNN model also achieved significantly lower errors than ANFIS and conventional empirical models for both PV technologies under all seasonal conditions. This work has applications in performance assessment and power prediction of solar power plant thereby supporting sustainable and reliable renewable energy deployment.