A Novel Renewable Power Generation Prediction Through Enhanced Artificial Orcas Assisted Ensemble Dilated Deep Learning Network

The different energy resource generation tends to have high-level variation, making the power supply complex for the end-users. Because of the intermittent nature, the variations occur by time, weather conditions, and output energy. Hence, this research aims to develop a new "Renewable Power Generation Prediction (RPGP)" model using Deep Learning (DL) to give the end user a reliable power supply. The data aggregation process initially accumulated the data in a normalized and structured format. Then, the data cleaning and scaling are performed to decrease the outliers and varying ranges of values. A higher-order statistical feature was attained from the cleaned and scaled data. This statistical feature was given to "Optimal Weight Computation Ensemble Dilated Deep Network (OWC-EDDNet)" to predict generated power. In this EDDLNet, networks such as "Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM), Deep Belief Networks (DBN), and Deep Neural Networks (DNN)" are employed to predict the renewable generated power. Finally, the prediction score attained from all deep networks is multiplied by the optimized weight to get the final prediction outcome, where the weights are optimally determined with the support of the Enhanced Artificial Orcas Algorithm (EAOA). The extensive empirical results were analyzed among traditional algorithms and prediction models to showcase the efficacy of the designed energy generation prediction scheme.