INTRODUCTION Accurate solar radiation projections can improve the efficiency of solar energy systems, optimize energy output, and help ensure the long-term viability of renewable energy projects. The purpose of this research is to develop a hybrid model that combines the strengths of both LSTM and GRU models to improve the quality of the forecast. Building on previous research, this study aims to improve understanding of Nigerian climate patterns and trends by identifying perceptions of solar radiation variability and change through Recurrent Neural Networks (RNN) such as Long Short-term Memory (LSTM) and Gated Recurrent Unit (GRU). MATERIALS AND METHODS This study examined historical climate data from the Nigerian Meteorological Agency's (NiMeT) database over a 31-year period. Monthly data were retrieved for seven cities in Nigeria: Ikeja, Sokoto, Maiduguri, Enugu, Ilorin, and Port Harcourt. The LSTM and GRU models were built using hyperparameter tuning to determine the best models. Both LSTM and GRU networks can accurately predict complex temporal relationships in data. CONCLUSIONS By successfully capturing convoluted temporal patterns, the panel hybrid RNN model—which combines LSTM and GRU—improves the accuracy of solar radiation predictions and is a beneficial tool for maximizing Nigeria's solar energy resources and assisting with sustainability initiatives. However, further improvements can be made by integrating real-time solar radiation data, expanding the model's application to larger datasets, and combining it with other machine learning techniques to enhance its predictive capabilities. Keywords: Hyperparameter tuning, Meteorological Data, Renewable Energy, Sequential Recurrent Neural Network, Solar Radiation Prediction, Time Series Forecasting
Alabi N. O., Ojenike O.T.