Advanced Prediction and Coated Solar Panel Performance Improvement Using Combined Long Short-Term Memory (LSTM) Architecture–Autoregressive Moving Average (ARMA) Technique

As solar energy has become a critical renewable resource, precise forecasting systems for photovoltaic (PV) solar panel power generation are becoming increasingly important. These panels were treated with hydrophobic coatings to increase their effectiveness and efficiency. In response to the increased demand for precise power forecasting, a new smart power prediction system was created specifically for coated PV solar panels. This novel approach used a hybrid model that included autoregressive moving average (ARMA) and long short-term memory (LSTM) approaches to successfully capture both short-term and long-term correlations in efficiency and output data. This method increased forecasting accuracy while addressing the constraints of older methods involving limitations in capturing both short-term and long-term dependencies in solar power generation data, reduced forecasting accuracy, and inefficiencies in feature extraction. Advanced feature extraction techniques, most notably the discrete wavelet transform (DWT), were used to identify important temporal and frequency patterns in solar insolation data. Following thorough testing and validation, the system achieved a very high accuracy of 98.3%, outperforming previous models by 2.3%. The deployment of this system resulted in considerable increases in PV efficiency, allowing for greater grid integration and energy management, ultimately contributing to a more sustainable energy future.