Scholarlink

<p>Electric vehicles (EVs) rely on lithium-ion batteries (LIBs) due to their high energy density, lightweight design, and long lifespan. Also, sudden dendrite formation in LIBs can lead to battery failure, reduced performance, safety hazards, and increased maintenance costs. But, none of the existing works provided timely alerts and focused on mitigation strategies during sudden dendrite formation in LIBs. To overcome this challenge, this paper introduces an integrated framework combining a Tsallis Sin-Swish ridge-based feed forward neural network (T2SR-FFNN), ZBellSin-fuzzy (ZBS-Fuzzy), and droop control system (DCS). Initially, the EV dataset is collected for the EV-power demand prediction. Then, the input data are preprocessed. From the preprocessed EV data, the features are gathered for predicting the power demand of EVs. On the LiB side, the dataset undergoes balancing, preprocessing, and feature extraction, thus enabling accurate state estimation using LM-EKF and effective cell balancing utilizing 2SCSB. LIB modeling with voltage, current, and temperature inputs supports real-time battery step-size prediction, whereas impedance measurement detects dendrite formation. A ZBS-Fuzzy system generates alerts, and if risks are detected, then a DCS is activated to mitigate hazards. Experimental validation demonstrates that the proposed framework enhances power demand prediction accuracy (i.e., 98.72%), ensures safe battery operation, and outperforms traditional methods in managing EV performance and safety.</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

Dendrite-based alert system and mitigation framework in lithium-ion EV batteries using T2SR-FFNN and ZBS-Fuzzy techniques

  • 期刊论文

  • 发表日期：

    2025年10月2日

查看全文
  • Puja Suresh Deokate,
  • Neeta Anilkumar Doshi

摘要

Electric vehicles (EVs) rely on lithium-ion batteries (LIBs) due to their high energy density, lightweight design, and long lifespan. Also, sudden dendrite formation in LIBs can lead to battery failure, reduced performance, safety hazards, and increased maintenance costs. But, none of the existing works provided timely alerts and focused on mitigation strategies during sudden dendrite formation in LIBs. To overcome this challenge, this paper introduces an integrated framework combining a Tsallis Sin-Swish ridge-based feed forward neural network (T2SR-FFNN), ZBellSin-fuzzy (ZBS-Fuzzy), and droop control system (DCS). Initially, the EV dataset is collected for the EV-power demand prediction. Then, the input data are preprocessed. From the preprocessed EV data, the features are gathered for predicting the power demand of EVs. On the LiB side, the dataset undergoes balancing, preprocessing, and feature extraction, thus enabling accurate state estimation using LM-EKF and effective cell balancing utilizing 2SCSB. LIB modeling with voltage, current, and temperature inputs supports real-time battery step-size prediction, whereas impedance measurement detects dendrite formation. A ZBS-Fuzzy system generates alerts, and if risks are detected, then a DCS is activated to mitigate hazards. Experimental validation demonstrates that the proposed framework enhances power demand prediction accuracy (i.e., 98.72%), ensures safe battery operation, and outperforms traditional methods in managing EV performance and safety.