Lévy Jump Nonlocal SPDE and BA-PINN Modeling for Battery Fracture and Thermal-Runaway Warning
Yongfang Zhu、Qing Xie、Jingli Jia
3. Lévy Jump Nonlocal SPDE and BA-PINN Modeling for Battery Fracture and Thermal-Runaway Warning
基本信息
- 标题:Lévy Jump Nonlocal SPDE and BA-PINN Modeling for Battery Fracture and Thermal-Runaway Warning
- DOI/URL:https://doi.org/10.3390/batteries12070249
- 期刊/会议/预印本平台:Batteries
- 发表日期:2026-07-12
- 作者和机构:Yongfang Zhu, Qing Xie, Jingli Jia;机构未能从可访问内容确认
- 通讯作者:未能确认
- 脚本评分:65
- 期刊指标核验:未核验
纳入理由和命中证据
- 摘要命中 lithium-ion batteries、degradation models、damage evolution、thermal-runaway warning。
- 方法包含 stochastic nonlocal fracture modeling 与 bifurcation-aware learning,并在 Oxford battery cycling records 与 Battery Failure Databank 上评估。
- 但核心任务是 fracture/thermal-runaway warning,不是 RUL/SOH/capacity fade 预测;因此仅作为“退化建模与预警方法观察项”。
创新点、主要内容、模型架构/方法流程
- 方法:Lévy-jump-driven nonlocal stochastic partial differential equation (SPDE) + Bifurcation-Aware Physics-Informed Neural Network (BA-PINN)。
- 框架耦合 fractional diffusion、peridynamic damage evolution、thermal feedback、state-space eigenvalue tracking 和 damage-variance monitoring。
- 创新点:将随机裂纹非局部建模与 bifurcation-aware 学习结合,用于临界故障/热失控预警。
输入特征与输出目标
- 输入特征:摘要未能完整确认;涉及 damage field、thermal feedback、cycling/abuse-test records。
- 输出目标:fracture damage field surrogate、thermal-runaway warning/risk;不是标准 RUL/SOH 输出。
数据来源/数据结构/电池类型/循环协议
- 数据来源:controlled synthetic fracture simulations、Oxford battery cycling records、Battery Failure Databank open-access abuse-test records。
- 电池类型、样本规模、循环协议:未能从可访问内容确认。
结果指标、Baseline/对比方法
- 指标:simulated fracture dataset 上 damage-field MSE 0.023 ± 0.002,SSIM 0.962 ± 0.006;thermal-runaway warning AUC-ROC 0.987 ± 0.004,平均 warning lead time 5.2 ± 0.2 h。
- Baseline/对比方法:未能从可访问内容确认。
泄漏风险/局限性/复现性
- 泄漏风险:中到信息不足。热失控事件数据通常样本少,需检查 event-level split、同源实验泄漏和 synthetic-to-real 泛化。
- 作者局限性:摘要明确说 direct experimental crack-map validation 不在当前数据范围内,需要 particle-resolved experiments 和更大 event-level thermal-runaway datasets。
- 复现性:信息不足;代码/数据处理细节未能确认。
对哥哥研究/落地的启发和后续动作
- 启发:BA-PINN 的 bifurcation/eigenvalue/damage-variance 思路可借鉴到“退化加速拐点检测”和安全预警,但不应直接作为寿命预测 baseline。
- 后续动作:低优先级下载全文;重点检查是否有可用于 SOH/RUL 的公开数据接口或特征工程。
查看关键词分类依据
- PINN / physics-informed · 标题包含「pinn」
Lévy Jump Nonlocal SPDE and BA-PINN Modeling for Battery Fracture and Thermal-Runaway Warning - Generic neural network / deep learning · 摘要包含「neural network」
el with a Bifurcation-Aware Physics-Informed Neural Network (BA-PINN). The framework couples fractional diffusion, peridynamic damage evolu - Hybrid model · 摘要包含「fusion」
A-PINN). The framework couples fractional diffusion, peridynamic damage evolution, thermal feedback, state-space eigenvalue trackin - Oxford · 摘要包含「oxford」
n controlled synthetic fracture simulations, Oxford battery cycling records, and open-access abuse-test records from the Battery Fa - Self-built / experimental dataset · 摘要包含「experimental」
and surrogate-learning evidence, with direct experimental crack-map validation remaining outside the present dataset scope. On the simula - Li-ion / LIB · 摘要包含「lithium-ion」
n major safety and durability challenges for lithium-ion batteries. Deterministic degradation models are limited in representing random