Hybrid Deep Learning-based Solar Irradiance Forecasting for Matarbari Coastal Region—The Next Generation Power Hub

Accurate prediction of solar irradiance is crucial for integrating solar power into power grids to enhance photovoltaic (PV) performance, particularly in regions with fluctuating and non-stationary weather conditions. This paper presents a hybrid deep learning model designed for Matarbari, Cox’s Bazar, Bangladesh, a coastal area characterized by highly variable and unpredictable climatic behavior. The proposed framework integrates an Artificial Neural Network (ANN) with a Bidirectional Long Short-Term Memory (BiLSTM) network to capture complex nonlinear relationships and temporal dependencies from high-resolution Solcast irradiance and meteorological data. By combining the feature-extraction capability of ANN with the sequence-learning strengths of BiLSTM, the hybrid model achieves improved forecasting accuracy and reliability. The model was trained and evaluated on multivariate coastal datasets and demonstrated strong generalization across a wide range of climate conditions. The proposed ANN-BILSTM hybrid model achieved exceptional performance on the test set, yielding a Coefficient of determination (R2) of 0.9768, a Root Mean Square Error of 0.0304, and a Mean Absolute Error of 0.0240 on scaled data. Overall, the proposed approach provides a robust data-driven forecasting solution that supports reliable solarpower integration, strengthens grid operational planning, and contributes to long-term energy management for Matarbari’s future power infrastructure.