Su Yong, Tian Shishun, Wu Hao, Li Xia
Guangdong Key Laboratory of Intelligent Information Processing, College of Electronics and Information Engineering, Shenzhen University, Shenzhen, Guangdong, China.
Sci Rep. 2025 Jul 8;15(1):24512. doi: 10.1038/s41598-025-08301-x.
The blooming population and advanced technology of electric vehicles (EVs) have promoted the wide studies on battery swapping station (BSS) models. However, current BSS researches only focus on isolated decision models, such as the dispatching model or charging schedule model, which cannot represent the realistic situation and obtain the optimal solution for both the EV drivers and BSS operators. In this paper, a bi-decision model for EV dispatch and BSS charging schedule problem is proposed to minimize the average extra time (ET) through the assigned BSS for EVs, and optimize the electricity cost, charging damage to batteries, and power load variance for BSSs, where the solution in the first decision is the pre-defined condition of the second decision model. Knowing that two models are both Non-deterministic Polynomial-time hard (NP-hard) problems, two types of evolutionary algorithms are proposed. In the first model, an adaptive tabu search (ATS) algorithm is proposed by formatting the EVs' ET, the number of batteries, and queuing EVs at BSSs. In the second model, a multi-objective particle swarm optimization (MOPSO) algorithm is proposed to obtain the Pareto set of the complicated scheduling problem. Experiments are carried out to investigate the viability of the bi-decision model by comparing it with rule-based strategies, such as nearest-in-range. Also, the waiting times in the first decision and the scheduling results are illustrated in the Gantt charts. Lastly, a comprehensive comparison between the proposed ATS algorithm and the MOPSO algorithm is presented to show the effectiveness and competitiveness.
电动汽车（EV）数量的不断增加和技术的进步推动了对电池更换站（BSS）模型的广泛研究。然而，目前的BSS研究仅关注孤立的决策模型，如调度模型或充电计划模型，这些模型无法代表实际情况，也无法为电动汽车驾驶员和BSS运营商获得最优解。本文提出了一种电动汽车调度和BSS充电计划问题的双决策模型，以通过为电动汽车分配BSS来最小化平均额外时间（ET），并优化BSS的电力成本、电池充电损耗和功率负荷方差，其中第一个决策的解是第二个决策模型的预定义条件。由于知道这两个模型都是非确定性多项式时间难（NP-hard）问题，因此提出了两种进化算法。在第一个模型中，通过对电动汽车的ET、电池数量和在BSS处排队的电动汽车进行格式化，提出了一种自适应禁忌搜索（ATS）算法。在第二个模型中，提出了一种多目标粒子群优化（MOPSO）算法来获得复杂调度问题的帕累托集。通过将双决策模型与基于规则的策略（如最近距离策略）进行比较，进行实验以研究该模型的可行性。此外，第一个决策中的等待时间和调度结果在甘特图中进行了说明。最后，对所提出的ATS算法和MOPSO算法进行了全面比较，以展示其有效性和竞争力。
