Artificial Intelligence-Driven Energy Management: A Strategy for Hybrid Electric Vehicle Trains Integrating Nondominated Sorting Dynamic Programming and Deep Learning

To enhance the adaptability of the energy management strategy (EMS) framework for trailers to varying vehicle parameters and driving conditions, this article proposes an artificial intelligence (AI)-driven EMS for hybrid electric vehicle (HEV) trains based on nondominated sorting dynamic programming (NSDP) and deep learning. First, an energy management model is developed using NSDP, from which optimal control sequences under different strategies are derived. In addition, a real-time energy management control approach based on long short-term memory (LSTM) networks is investigated. The training dataset is generated using optimization results obtained from the NSDP-based model, and the optimal power distribution sequences are extracted to construct the real-time control strategy. Finally, simulation experiments conducted on actual logistics transportation routes demonstrate that, compared with a rule-based strategy, the LSTM-based strategy reduces equivalent fuel consumption by 4.37% and battery degradation by 47.49%, thereby validating the adaptability of the proposed EMS to diverse parameters and operating conditions.