Quantum Reinforcement Learning for real-time optimization in Electric Vehicle charging systems

Abstract

The rapid growth of electric vehicles (EVs) presents new challenges for EV charging scheduling, particularly due to the unpredictable nature of charging demand and the dynamic availability of resources. Currently, Deep Reinforcement Learning (DRL) has become a critical technology for improving scheduling efficiency. At the same time, advancements in quantum computing have led to Quantum Neural Networks (QNNs), which use the superposition states of quantum bits for more efficient information encoding. Building on these advancements, this study explores Quantum Reinforcement Learning (QRL) for EV charging systems. We propose a method called QRL-based Electric Vehicle Charging Scheduling (Q-EVCS) to optimize charging resource allocation based on real-time user demand. This approach aims to reduce average charging service times, increase the service success rate, and lower operational costs. We provide the detailed design and implementation of our approach, and our experimental results demonstrate that Q-EVCS maintains performance levels comparable to the DRL-based method while significantly reducing the number of model parameters.