Scholarlink

With the growing adoption of electric vehicles (EVs), precise State of Charge (SOC) estimation is essential for optimizing battery management and ensuring reliable vehicle performance. This chapter provides a detailed analysis of SOC estimation in EVs, focusing on the application of Artificial Neural Networks (ANNs) and comparing their performance with the Extended Kalman Filter (EKF) and coulomb counting methods. The study is conducted using the Tesla Model S across two distinct driving conditions: the Highway (HW) driving cycle and the City (CY) driving cycle, providing a comprehensive evaluation across varied operational environments. The results show that the ANN method consistently outperforms the EKF, achieving up to 27.4% lower Root Mean Square Error (RMSE) in the highway driving cycle, indicating superior accuracy in SOC estimation. Even in the city driving cycle, the ANN reduces RMSE by up to 3.8%, demonstrating its robustness across various temperatures and driving conditions. These findings underscore the potential of ANN for SOC estimation, making it a viable and promising alternative for EVs.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Performance Evaluation of Artificial Neural Networks for Electric Vehicle State of Charge Estimation Across Different Driving Cycles

  • 章节

  • 发表日期：

    2025年

查看全文
  • Battula Amrutha Raju,
  • Dushyant Sharma,
  • Surender Reddy Salkuti

摘要

With the growing adoption of electric vehicles (EVs), precise State of Charge (SOC) estimation is essential for optimizing battery management and ensuring reliable vehicle performance. This chapter provides a detailed analysis of SOC estimation in EVs, focusing on the application of Artificial Neural Networks (ANNs) and comparing their performance with the Extended Kalman Filter (EKF) and coulomb counting methods. The study is conducted using the Tesla Model S across two distinct driving conditions: the Highway (HW) driving cycle and the City (CY) driving cycle, providing a comprehensive evaluation across varied operational environments. The results show that the ANN method consistently outperforms the EKF, achieving up to 27.4% lower Root Mean Square Error (RMSE) in the highway driving cycle, indicating superior accuracy in SOC estimation. Even in the city driving cycle, the ANN reduces RMSE by up to 3.8%, demonstrating its robustness across various temperatures and driving conditions. These findings underscore the potential of ANN for SOC estimation, making it a viable and promising alternative for EVs.