A data-driven approach for collaborative optimization of large-scale electric vehicles considering energy consumption uncertainty

With the explosive growth of electric vehicles (EVs), it is an urgent task to incorporate low-carbon EVs with advanced optimization strategies to achieve orderly charging–discharging and economical operations for EVs. Nevertheless, current charging–discharging optimization strategies for EVs may be impractical since their energy consumption uncertainty and the driving trip cost are generally not considered. Therefore, a collaborative optimization model for large-scale EV charging–discharging with energy consumption uncertainty in this paper is proposed to simultaneously maximize passenger revenue and reduce the costs of the driving, charging–discharging, and battery depletion. Subsequently, a data-driven approach is developed to tackle the model. In this approach, an uncertainty predictor based on wavelet transform, deep deterministic policy gradient, and quantile regression is first applied to estimate the energy consumption uncertainty. Then, an adaptive learning rate firefly algorithm is presented to identify the most satisfactory solution for the optimization model. Finally, taking the actual data of 300 EVs in a city in China as case studies, the simulation results reveal that the proposed method is effective and has high application significance. •A new collaborative optimization model for EV charging–discharging is formulated.•A deep reinforcement learning method for energy consumption prediction is proposed.•The impact of driving behavior on energy consumption prediction is studied.•A new adaptive learning rate Firefly algorithm is presented and applied.•The effectiveness of the proposed methodology is verified by simulations of 300 EVs.