Deep Reinforcement Learning-Based Navigation Strategy for a Mobile Charging Station in a Dynamic Environment

Finding effective charging techniques is necessary as the number of electric vehicles (EVs) on the transportation network rises. On this basis, mobile charging stations (MCSs) play a vital role for supplying EV charging demands. However, it is difficult to model real time variables such as traffic conditions on the transportation network and dynamic charging demands from EV owners. This study proposes a novel deep reinforcement learning (DRL) based MCS navigation technique to handle the real time randomness. Hence, an energy and time aware MDP formulation is constructed with objective of minimizing the total travel cost and total time cost of MCS. Moreover, the proximal policy optimization (PPO) algorithm is utilized to train the MCS agent. The effectiveness of the developed model is tested with randomness on transportation and charging demands. Numerical results show that the MCS agent provides the optimal route and battery usage on the network.