Effective Charging Scheduling of Electric Vehicles Using a Hybrid Deep Learning Network

Abstract

Electric vehicles (EVs) are developed by diverse industries as a substitute for vehicles with internal combustion engines, with many compensations that are environment-friendly. The amount of EVs is likely to rise fast in the approaching ages. However, uncoordinated vehicle charging may significantly stress the power grid. The main objective of the devised model is to minimize the charging time and waiting time for EVs by distributing equal power resources. Therefore, an energy-aware multi-objective system in a cloud-internet of things (IoT)-based electric vehicular network for a priority-based charge-scheduling scheme is proposed here and established as follows. Initially, the network with the EV location as well as the charge station (CS) location is simulated. Then, the charging planning is performed by determining the CS selection using the fractional spotted hyena jellyfish optimization (FSHJSO) considering a multi-objective function. Subsequently, the charge scheduling is performed using the established hybrid deep learning (DL) approach namely MobileNet neural network (MNN-Net) based on various objectives. The integration of MobileNet with deep neural network (DNN) forms the MNN-Net. By employing deep neuro-fuzzy network (DNFN), the power prediction is done. The efficiency of the developed MNN-Net is validated with some methods and achieved superior performance with an average waiting time of 11.796 s, distance 0.067 m, available power 53.657 W and number of EVs charged 63.