Reinforcement Learning Based Algorithm for the Maximization of EV Charging Station Revenue

TL;DR

This paper introduces a reinforcement learning algorithm that dynamically optimizes EV charging station revenue by adapting to customer trends and renewable energy sources, validated through computer simulations.

Contribution

It proposes a novel Q-learning based method for maximizing EV station revenue that adapts to changing customer patterns and renewable energy availability.

Findings

The algorithm effectively increases revenue in simulated environments.

It adapts to shifts in customer behavior and energy supply.

Simulation results confirm the model's utility.

Abstract

This paper presents an online reinforcement learning based application which increases the revenue of one particular electric vehicles (EV) station, connected to a renewable source of energy. Moreover, the proposed application adapts to changes in the trends of the station's average number of customers and their types. Most of the parameters in the model are simulated stochastically and the algorithm used is a Q-learning algorithm. A computer simulation was implemented which demonstrates and confirms the utility of the model.