摘要
在电动汽车(EV)换电模式中,换电站的电池充放电策略对运营商、配电网和EV换电成本十分重要。在实际运营中,一个关键问题是实时决定充放电的电池数量和功率。提出一种基于门控循环单元(GRU)的实时调度方法,可以在线做出充放电决策。采用建立的换电站充电模型对大量场景进行优化,并通过数据预处理将优化结果转换为包含历史信息和预测信息的输入输出样本对。使用GRU学习生成的样本,自主建立不同时间和特征与充电决策之间的映射。通过设计输出结果标准化方法,使得最终输出符合物理逻辑。训练完毕的神经网络即可在线部署,求解时间仅需若干毫秒。仿真表明,所提方法可以显著降低充电成本并提高换电站服务质量。
In the battery swapping mode of the electric vehicle(EV),the battery charging strategy of battery swapping station is significant to the swapping operator,distribution network and EV swapping cost.In actual operation,a critical issue is to determine the number and power of batteries to charge and discharge in real-time.This paper proposes a real-time scheduling method based on gate recurrent unit(GRU),which can make charging and discharging decisions online.The battery swapping station charging model is established,and the optimization results are converted into input samples containing historical and predictive information through data preprocessing.GRU is utilized to learn the generated samples and automatically establish mappings from different times and features to charging decisions.By designing the output standardization method,the final output conforms to the physical logic.The trained neural network can be deployed online to solve problems in milliseconds.Simulation results demonstrate that the proposed method can significantly reduce the operating cost and improve the service quality of the battery swapping station.
作者
张吉波
王胜生
王子奇
ZHANG Jibo;WANG Shengsheng;WANG Ziqi(Zhangye Power Supply Company of State Grid Gansu Electric Power Company,Zhangye 734000,China;School of Electrical and Electronic Engineering,North China Electric Power University,Baoding 071003,China)
出处
《微型电脑应用》
2024年第11期263-267,共5页
Microcomputer Applications
关键词
门控循环单元
深度学习
电动汽车
换电
实时调度
gate recurrent unit
deep learning
electric vehicle
battery swapping
real-time scheduling
