# Personalized Dynamic Pricing Policy for Electric Vehicles: Reinforcement learning approach

**Authors:** Sangjun Bae, Balazs Kulcsar, Sebastien Gros  
**arXiv:** [2401.00661](https://arxiv.org/abs/2401.00661)  
**Date:** 2024-01-01  
**Summary page:** https://www.summarizepaper.com/en/arxiv-id/2401.00661v1/

## Key points

- Study explores impact of information technology on electricity price competition within fast-electric vehicle charging station market
- Focuses on self-interested EV users maximizing utilities when selecting fast-EVCS for charging
- Proposes personalized dynamic pricing policy (PeDP) using reinforcement learning approach to optimize revenue generation
- Introduces multiple fast-EVCS competing simulation environment modeling EV users' behavior through game-based charging station selection model
- Highlights significance of waiting time in EV charging market and evaluates PeDP performance considering different types of information from service provider's standpoint
- Discusses potential misuse scenarios where privacy-preserved information sharing can be exploited by AI-driven PeDP systems from customer perspective

## Summary

The study "Personalized Dynamic Pricing Policy for Electric Vehicles: Reinforcement learning approach" by Sangjun Bae, Balazs Kulcsar, and Sebastien Gros explores the impact of information technology on electricity price competition within the fast-electric vehicle charging station (fast-EVCS) market. With the increasing number of fast-EVCSs, utilizing public and privacy-preserved information becomes crucial in shaping competitive dynamics. The research focuses on self-interested electric vehicle (EV) users who aim to maximize their utilities when selecting a fast-EVCS for charging based on factors such as electricity price, estimated waiting time, and state of charge. Using a reinforcement learning approach, the authors propose a personalized dynamic pricing policy (PeDP) to optimize revenue generation for fast-EVCSs. They also introduce a multiple fast-EVCS competing simulation environment that models EV users' behavior through a game-based charging station selection model with a monetary utility function. Through numerical simulations, the study highlights the significance of waiting time in the EV charging market and evaluates the performance of PeDP while considering different types of information from the service provider's standpoint. It also sheds light on potential misuse scenarios where privacy-preserved information sharing can be exploited by artificial intelligence-driven PeDP systems from a customer perspective. Overall, this study provides valuable insights into personalized dynamic pricing strategies for fast-EVCSs using cutting-edge reinforcement learning techniques and emphasizes the importance of considering various factors in optimizing revenue generation within this evolving market landscape.