Hybrid Metaheuristics with Sparse-Trained Deep Learning for Sustainable Electric Vehicle Charging Demand Forecasting

Electric vehicles (EVs) are generally been penetrated to power schemes due to their huge advantages over fossil fuel vehicles, viz., superior security of energy and lesser emission of greenhouse gases. With improving the EV penetration, the energy management of power grids is developing further comp...

Full description

Saved in:
Bibliographic Details
Main Authors: Alkhafaij, Mahdi Abdulkhudur, Al-luhiby, Hussein A., Al-Hameed, Mazin Riyadh, Saleem, Munqith, Habelalmateen, Mohammed I., Mohammed, E. Ali
Format: Conference Proceeding
Language:English
Subjects:
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Electric vehicles (EVs) are generally been penetrated to power schemes due to their huge advantages over fossil fuel vehicles, viz., superior security of energy and lesser emission of greenhouse gases. With improving the EV penetration, the energy management of power grids is developing further complicate and challenges caused by outcomes of EVs on market prices and current consumption. Therefore, correct EV charging load demand prediction is most important concern from the power grid viewpoint. For reducing a system computation burden and to take the difficult controller realtime execution, the deep learning (DL) approaches are utilized ever more to enhance the extremely non-linear method forecast efficiency. Therefore, this study introduces a new Hybrid Metaheuristics with Sparse-Trained Deep Learning for Sustainable Electric Vehicle Charging Demand Forecasting (HMSDL-EVCDF) technique. The goal of the HMSDL-EVCDF technique is to investigate the input time series data for prediction process. In the presented HMSDL-EVCDF technique, data decomposition process takes place by the use of empirical mode decomposition (EMD) for maintaining the features. For prediction process, the HMSDL-EVCDF technique uses sparse trained recurrent neural network (STRNN) model. To enhance the predictive outcomes of the STRNN model, hybrid magnetic optimization with particle swarm optimization (HMOPSO) algorithm is used as hyperparameter optimizer. The experimental validation of the HMSDL-EVCDF technique is tested on different case studies and the outcomes are examined with respect to various measures. The extensive comparison study portrayed the improved performance of the HMSDL-EVCDF system over other recent approaches.
ISSN:2831-753X
DOI:10.1109/IICETA57613.2023.10351350