Incentive-Based Distributed Scheduling of Electric Vehicle Charging Under Uncertainty

We consider a power system with
$N$
charging stations, each of which has local renewable generation and electric vehicle (EV) arrivals with deadline constraints for their requests. To reduce the adverse impact brought by uncoordinated charging, the utility company requires to adopt advanced load management to control the load supplied to stations but without collecting many details. The major challenge is to incentivize them to coordinate charging for its system-wide optimization while meeting all charging requests, regarding inherent uncertainties. We model a game that aims to minimize the total electricity cost at the utility company meanwhile maximize the payoff of each station. We combine
$\epsilon$
-Nash equilibrium and Lyapunov optimization to design a low-complexity distributed scheduling scheme of EV charging. It achieves at most
$O(1/V)$
more than the optimal cost where
$V$
is a controllable parameter that balances between the cost and average fulfillment ratio of charging requests. The payoff of each station is around
$O(1/V)$
less than the optimal payoff if using a small
$V$
. Simulation results under real-world traces show that, compared with a state-of-the-art scheduling algorithm, our algorithm reduces the peak-to-average ratio and charging cost, respectively, by 94.32% and 44.21%, and increases the payoff by 52.27% averagely.