A Game-theoretic Approach for Dynamic Service Scheduling at Charging Facilities

TL;DR

This paper introduces a dynamic, game-theoretic scheduling approach for EV charging stations that accounts for demand uncertainties, aiming to optimize costs and improve user experience in real-time.

Contribution

It develops a novel dynamic programming model combined with a generalized Nash equilibrium and Monte Carlo tree search to efficiently handle EV charging uncertainties.

Findings

The method effectively minimizes travel, waiting, and charging costs.

Numerical experiments show high solution quality and computational efficiency.

The approach supports EV adoption and environmental sustainability.

Abstract

Electric vehicle (EV) charging patterns are highly uncertain in both location, time, and duration particularly in association with the predicted high demand for electric mobility in the future. An EV can be charged at home, at charging stations near highway ramps, or on parking lots next to office buildings, shops, airports, among other locations. Charging time and duration can be fixed and continuous or flexible and intermittent. EV user preferences of charging services depend on many factors (e.g., charging prices, choice of destinations), causing EV charging patterns to shift in real-time. Hence, there is a need for a highly flexible EV charging network to support the rapid adoption of the technology. This study presents a dynamic scheduling scheme for EV charging facilities considering uncertainties in charging demand, charger availability, and charging rate. The problem is formulated as a dynamic programming model that minimizes the travel and waiting costs and charging expenses while penalizing overcharging attempts. An integrated generalized Nash equilibrium technique is introduced to solve the problem that incorporates a Monte Carlo tree search algorithm to efficiently capture the uncertainties and approximate the value function of the dynamic program. Numerical experiments on hypothetical and real-world networks confirm the solution quality and computational efficiency of the proposed methodology. This study will promote EV adoption and support environmental sustainability by helping users lower the charging spot search burden via a real-time, user-adaptive optimizer. Stakeholders can retrieve charger utilization and pricing data and get feedback on their charging network policies.