Enriching redistribution of power in EV Charging Stations through Deep learning

Abstract

Extensive population growth and increased usage of electric devices are leading to unprecedented levels of electricity consumption. At times, this consumption can surge significantly during peak hours, leading to power outages in certain areas and occasionally even causing grid failures. Fast electric vehicle charging stations often use more power depending on the number of vehicles charging at a given time. A grid failure during peak hours could significantly impact the overall business of electric vehicle charging station owners, potentially leading to a decrease in the use of electric vehicles among consumers. This could potentially worsen air pollution by increasing the use of fossil fuel-powered vehicles. Therefore, it is crucial to regulate power at charging stations according to slot requirements for efficient charging, ensuring that only a larger number of electric vehicles can be successfully charged during periods of moderate load. Hence, to dynamic redistribution of power among charging ports within a station based on current demand and battery state-of-charge (SoC) is the actually task  which can ease the load of electricity  properly. This approach not only minimizes user wait times but also maximizes energy utilization, thereby increasing the popularity of electric vehicles among consumers and contributing to the reduction of pollution. The deep learning model  Deep belief neural networks, in conjunction with k-nearest neighbor models and decision trees, function as catalysts to determine the dynamic redistribution of power among the slots, taking into account the port's state of charge, thereby enhancing the process. Further, root mean square error measurement shows that the designed model yields an exceptionally lower error rate in the redistribution of power within the station based on current demand and battery state-of-charge.