Energy is the most appropriate method to fill the hole of social, financial, and ecological factors for improving human growth across the world. Renewable energy prediction is of paramount importance for reliable operation and efficient management of power systems in a continuously changing climate. Nevertheless, there are still deficits in the accuracy of existing forecaster model, poor generalization, and ineffective optimization results of parameters. To cope with those difficulties, we present in this study a new global renewable energy forecasting model using hybrid deep learning and two‐tier optimization technics. Our proposed model integrates wisdom of features, attention‐guided spatiotemporal learning, and automatic determination of hyperparameters to ensure robust and accurate prediction. Through extensive experiments done on a benchmark global renewable energy dataset, the experimental results show that our proposed approach outperforms other state‐of‐the‐art methods in terms of prediction accuracy, stability, and generalization ability. For the past few years, developments in technology, computing power, and artificial intelligence (AI) have delivered an effective method for tackling numerous urban computing issues such as short‐term renewable energy prediction. This study presents a novel global renewable energy forecasting using a hybrid deep learning and two‐Tier optimization models (GREF‐HDLTOM). The proposed GREF‐HDLTOM model’s main intention is to enhance the prediction model of renewable energy using advanced DL and optimization algorithms. At first, the data preprocessing stage contains input scaling using a standard scaler and output scaling using MaxAbsScaler for converting the categorical data into a numerical format. For the feature selection process, the proposed GREF‐HDLTOM model designs an African vulture optimization algorithm (AVOA). This study introduces a new global renewable energy forecasting model using hybrid deep learning and two‐tier optimization method. Its components consist of AVOA for feature selection and attention‐convolutional gated recurrent neural networks (A‐CGRNNs) for spatiotemporal predictions, while improved northern goshawk optimization (INGO) is employed to search optimal hyperparameters. Experiments show that the new method significantly outperforms existing methods. Furthermore, the hybrid of A‐CGRNN has been deployed for the prediction process. At last, the INGO algorithm adjusts the hyperparameter values of the A‐CGRNN model optimally and outcomes in greater prediction performance. The experimental evaluation of the GREF‐HDLTOM system occurs utilizing a benchmark database. The simulation outcomes indicated the enhanced performance of the GREF‐HDLTOM system compared to existing approaches.
Mohammed et al. (Thu,) studied this question.