汽车工程 ›› 2024, Vol. 46 ›› Issue (11): 2059-2067.doi: 10.19562/j.chinasae.qcgc.2024.11.012

• • 上一篇    下一篇

基于Transformer的纯电动汽车充电时间预测

胡杰1,2,3(),陈琳1,2,3,王志红1,2,3,卿海华1,2,3,王浩杰1,2,3   

  1. 1.武汉理工大学,现代汽车零部件技术湖北省重点实验室,武汉 430070
    2.武汉理工大学,汽车零部件技术湖北省协同创新中心,武汉 430070
    3.湖北省新能源与智能网联车工程技术研究中心,武汉 430070
  • 收稿日期:2024-03-30 修回日期:2024-05-16 出版日期:2024-11-25 发布日期:2024-11-22
  • 通讯作者: 胡杰 E-mail:auto_hj@163.com
  • 基金资助:
    广西省科技重大专项(2023AA03009)

Transformer-Based Prediction of Charging Time for Pure Electric Vehicles

Jie Hu1,2,3(),Lin Chen1,2,3,Zhihong Wang1,2,3,Haihua Qing1,2,3,Haojie Wang1,2,3   

  1. 1.Wuhan University of Technology,Hubei Key Laboratory of Modern Auto Parts Technology,Wuhan 430070
    2.Wuhan University of Technology,Auto Parts Technology Hubei Collaborative Innovation Center,Wuhan 430070
    3.Hubei Technology Research Center of New Energy and Intelligent Connected Vehicle Engineering,Wuhan 430070
  • Received:2024-03-30 Revised:2024-05-16 Online:2024-11-25 Published:2024-11-22
  • Contact: Jie Hu E-mail:auto_hj@163.com

摘要:

纯电动汽车充电时间的安排是车主日常生活中至关重要的环节,直接影响车主出行的便利度和舒适体验。然而,目前仍然面临充电桩资源不足、充电须提前规划等挑战,为解决车主因车辆电量不足而无法立即用车的问题,提出一种基于 Transformer 模型的充电时间预测解决方案,帮助车主更好地规划日常行程。为了更好地了解电池性能衰减程度和容量损失情况,采用容量法评估电池健康状态,并分析驾驶人的充电行为,对电池充电行为特征进行构建。使用Savitzky-Golay 滤波器对表征电池衰减的特征进行平滑处理,并进行累积变换,使特征能更全面地表征电池信息;再耦合皮尔逊相关系数和 LASSO(least absolute shrinkage and selection operator)回归算法二次筛选得到最优特征集。最后,利用 Transformer 模型的超强注意力机制,对充电时间进行预测。通过实验数据验证,此方案可以准确且快速地预测纯电动汽车的充电时间,决定系数达到0.999,运行时间为156 ms。

关键词: 电动汽车, 充电时间, 数据驱动, Transformer

Abstract:

The arrangement of charging time for pure electric vehicles is a crucial part of the daily life of car owners, directly affecting the convenience and comfortable experience of their travel. However, there are still challenges such as insufficient charging station resources and the need for advanced planning for charging. To solve the problem of car owners being unable to use the vehicle immediately due to insufficient battery, a charging time prediction solution based on the Transformer model is proposed to help car owners better plan their daily itinerary. In order to better understand the degree of battery performance degradation and capacity loss, the capacity method is used to evaluate the health status of batteries, and the charging behavior of drivers is analyzed to construct the characteristics of battery charging behavior. Savitzky Golay filter is used to smooth out the features representing battery attenuation and perform cumulative transformation, so that the features can more comprehensively represent battery information. Then the Pearson correlation coefficient and LASSO (Least Absolute Shrinkage and Selection Operator) regression algorithm are coupled to obtain the optimal feature set through secondary screening. Finally, using the Transformer model's strong attention mechanism, the charging time is predicted. Through experimental data verification, this scheme can accurately and quickly predict the charging time of pure electric vehicles, with a determination coefficient of 0.999 and a running speed of 156 ms.

Key words: electric vehicles, charging time, data driven, Transformer