Electric vehicle optimum charging-discharging scheduling with dynamic pricing employing multi agent deep neural network

• Published in: Computers & electrical engineering Vol. 105; p. 108555
• Main Authors: Aljafari, Belqasem, Jeyaraj, Pandia Rajan, Kathiresan, Aravind Chellachi, Thanikanti, Sudhakar Babu
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.01.2023
• Subjects: Deep neural network; Dynamic pricing; EV Charging scheduling; Queuing model; Smart grid; Dynamic pricing; Deep neural network; EV Charging scheduling; Queuing model; Smart grid
• ISSN: 0045-7906; 1879-0755
• DOI: 10.1016/j.compeleceng.2022.108555

Electric Vehicles (EVs) are environmentally friendly. Extensive progress makes EVs popularly deployed and adopted. Once EVs are connected to the smart grid, EVs can act as both variable load and energy supply systems. One major challenge in EV deployment is the management of charging stations with minimum waiting time and reduced EV customer electricity prices. Considering dynamic pricing and various EV features could provide optimum scheduling. To address this issue, we proposed dynamic pricing and optimized scheduling as constrained by a Markov decision process. The solution is obtained by a novel Multi-Agent Deep Neural Network (MADNN). A numerical experiment was conducted with real-time data using the Nissan Leaf model EV. The proposed MADNN uses queuing model and obtained the highest saving rate of 18.45% and an average profit of 340.5 $/kWh with a network convergence time of 520 s. This obtained result validates the effectiveness of the proposed EV scheduling algorithm.