Solar Energy Forecasting using Optimized Attention-based Bidirectional Long Short-Term Memory

Solar energy is abundantly available in the environment and is beneficial for preserving the environment, since they are regenerative and hold significant promise for the future. Forecasting accurate solar energy plays a major part in enhancing the determination of solar power plants, while also decreasing reliance on fossil fuels for economic and social progress. In the domain of deep learning (DL), the prediction of solar resources has transitioned from traditional statistical methods to the implementation of advanced DL models. This work introduces a model of "optimized attention-bidirectional long short-term memory" (OA-Bi-LSTM) for predicting solar energy. Here, the hyper-parameters of the BiLSTM are optimized by the manta ray optimization (MRO). The proposed model predicts solar energy in an efficient manner with respect to the different measures and achieves better MAE (0.15) and RMSE (0.21) values respectively.