Scholarlink

<p>The rapid adoption of electric vehicles (EVs) has highlighted the critical importance of reliable battery management systems to ensure their safety and performance. In particular, electric vehicle batteries are prone to faults that can result in thermal runaway, endangering both vehicle and passenger safety. Existing fault diagnosis methods face challenges in accuracy and adaptability due to the complexity of battery systems and varying environmental conditions. This study introduces a novel fault diagnosis method that integrates advanced cyber-physical systems (CPS) to enhance the precision and reliability of electric vehicle battery fault detection. By combining real-time battery data with historical information, this approach predicts environmental changes and adapts the battery model accordingly. Specifically, Long Short-Term Memory Networks (LSTMN) and Back Propagation Neural Networks (BPNN) are used to predict temperature variations under different conditions, leading to a self-updating battery model. This method significantly improves fault warning times by an average of 57 seconds, reduces misdiagnosis by 11.1%, and decreases diagnostic failure rates by 8.4%, when compared to traditional model-based approaches. Validation with real-world data from electric buses demonstrates the effectiveness of this novel approach in real-time fault prediction and diagnosis, offering a significant advancement in electric vehicle safety and reliability.</p>

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

From Environmental Perception to Intelligent Battery Management: A Novel Real-Time Fault Early Warning Method for Onboard Power Batteries in a Cyber-Physical System

  • 期刊论文

  • 发表日期：

    2025年11月21日

查看全文
  • Jingxuan Zhang

摘要

The rapid adoption of electric vehicles (EVs) has highlighted the critical importance of reliable battery management systems to ensure their safety and performance. In particular, electric vehicle batteries are prone to faults that can result in thermal runaway, endangering both vehicle and passenger safety. Existing fault diagnosis methods face challenges in accuracy and adaptability due to the complexity of battery systems and varying environmental conditions. This study introduces a novel fault diagnosis method that integrates advanced cyber-physical systems (CPS) to enhance the precision and reliability of electric vehicle battery fault detection. By combining real-time battery data with historical information, this approach predicts environmental changes and adapts the battery model accordingly. Specifically, Long Short-Term Memory Networks (LSTMN) and Back Propagation Neural Networks (BPNN) are used to predict temperature variations under different conditions, leading to a self-updating battery model. This method significantly improves fault warning times by an average of 57 seconds, reduces misdiagnosis by 11.1%, and decreases diagnostic failure rates by 8.4%, when compared to traditional model-based approaches. Validation with real-world data from electric buses demonstrates the effectiveness of this novel approach in real-time fault prediction and diagnosis, offering a significant advancement in electric vehicle safety and reliability.