A Deep CNN-LSTM Model Tuned by PPSO for Spatiotemporal Solar GHI Forecasting

Accurately forecasting solar global horizontal irradiance (GHI) is a challenging and crucial task for both photovoltaic power plants and grid operators. This paper proposes a novel deep learning model based on a convolutional neural network (CNN) and long short-term memory (LSTM) for 24-hour-ahead solar GHI forecasting, utilizing spatiotemporal information. The proposed model can capture both the spatial and temporal dependencies of solar GHI data from multiple locations and time windows for input data. To optimize the parameters/hyperparameters of the model, such as the number of layers, filters, and neurons, a phasor particle swarm optimization (PPSO) algorithm is employed, which is a variant of the particle swarm optimization (PSO) algorithm and utilizes complex numbers to represent the particles and their velocities. The PPSO algorithm is efficient in exploring the search space and avoiding local optima. The proposed model is evaluated using data from different countries and compared with several baseline methods. The results show that the proposed model outperforms the baseline methods in terms of accuracy and robustness.