Chen Kaizhe, Liu Jin, Lyu Wenjing, Wang Tianyuan, Wen Jinxi
School of Education, Beijing Institute of Technology, Beijing, China.
Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology Cambridge, American, United States.
Sci Rep. 2025 Jul 1;15(1):21451. doi: 10.1038/s41598-025-05953-7.
The rapid advancement of Artificial Intelligence (AI) has led to a profound transformation in the transportation industry, particularly in driving the shift toward carbon neutrality and electrification. AI has proven to be a key enabler in formulating innovative strategies for optimizing electric vehicle (EV) fleets, thus advancing transportation services. While extensive research has been conducted on AI's role in transportation innovation, there remains a significant gap in empirical studies focusing on optimizing the charging behavior of operational EV fleets, particularly within ride-hailing services. The rise of ride-hailing services has revolutionized the transportation landscape, and their transition to EV fleets presents a major opportunity. The integration of AI to optimize the operations of these EV ride-hailing fleets could substantially help achieve the dual objectives of reducing charging costs and simultaneously lowering carbon emissions. Therefore, this research develops a Neural Network (NN) trained with the Adaptive Moment Estimation (Adam) algorithm, based on 2.14 million charging events. The goal is to analyze current charging behaviors and evaluate the impact of key variables on costs and emissions, providing data-driven insights for potential improvements, thus addressing a critical research gap. The novelty of this study lies in its novel combination of deep learning algorithms with large-scale real-world charging data, proposing a new method for optimizing EV ride-hailing charging behavior, and providing practical solutions for promoting electric vehicle adoption and achieving low-carbon transportation.
人工智能（AI）的快速发展给交通运输行业带来了深刻变革，尤其是推动了向碳中和与电气化的转变。事实证明，人工智能是制定优化电动汽车（EV）车队创新策略的关键推动因素，从而推动了交通运输服务的发展。虽然已经对人工智能在交通创新中的作用进行了广泛研究，但在针对优化运营中的电动汽车车队（尤其是在网约车服务中）的充电行为的实证研究方面，仍存在重大差距。网约车服务的兴起彻底改变了交通格局，其向电动汽车车队的转型带来了重大机遇。整合人工智能以优化这些电动网约车车队的运营，有助于大幅实现降低充电成本和同时减少碳排放的双重目标。因此，本研究基于214万个充电事件，开发了一种采用自适应矩估计（Adam）算法训练的神经网络（NN）。目标是分析当前的充电行为，评估关键变量对成本和排放的影响，提供数据驱动的见解以实现潜在改进，从而填补一个关键的研究空白。本研究新颖之处在于将深度学习算法与大规模实际充电数据进行了新颖的结合，提出了一种优化电动网约车充电行为的新方法，并为促进电动汽车的采用和实现低碳交通提供了切实可行的解决方案。
