Toward Trustworthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity Validation Framework from Laboratory Cells to Real-World Packs
Stefan Rizanov、Anna Stoynova、Georgy Mihov
3. Toward Trustworthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity Validation Framework from Laboratory Cells to Real-World Packs
论文基本信息
- 标题：Toward Trustworthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity Validation Framework from Laboratory Cells to Real-World Packs
- DOI/URL：https://doi.org/10.3390/batteries12070255
- 期刊/会议/预印本平台：Batteries
- 发表日期：2026-07-15（Crossref published-online）
- 作者和机构：Stefan Rizanov（Technical University of Sofia, Department of Electronics）；Anna Stoynova（Technical University of Sofia, Department of Microelectronics）；Georgy Mihov（Technical University of Sofia, Department of Electronics）
- 通讯作者：未能确认（MDPI 页面/PDF 抓取失败，Crossref 未给通讯作者字段）
- 脚本评分：76
- 期刊指标核验：未核验 JCR/JIF/JCI/CiteScore/SJR；脚本标注为 specialized battery journal。
- 可信度：摘要/元数据级。已尝试 PDF、MDPI 页面、DOI，web_extract 失败；Crossref API 可确认题名、作者、期刊、日期和摘要。
纳入理由和命中证据
- 纳入理由：直接讨论 lithium-ion battery diagnostics/prognosis/management 中 SOH 和 RUL 估计的可信性与迁移性。
- 命中证据：摘要写明 reliable SOH and RUL estimation is essential；synthesizes model-based, machine learning, deep learning, transfer learning, physics-informed, impedance-based, thermographic, relaxation-based, and digital twin approaches。
- 与普通材料论文区别：核心对象是估计器评价和部署验证，不是材料制备或单一循环性能。
创新点
- 把部署前证据要求形式化为 multi-fidelity validation framework：synthetic cells → controlled aging → module/pack testing → fleet shadow operation → closed-loop safety-governed deployment。
- 强调 worst-domain error、uncertainty calibration、warning risk、computational feasibility，而不是单一平均误差。
主要内容
- 对 176 篇去重分类科学文献进行综述，并结合 supplementary evidence audit workbook。
- 讨论实验室 cell 数据与真实 pack/fleet 数据之间的 domain shift、cross-fidelity degradation、pack heterogeneity。
- 主张把 battery health intelligence 评价为 evidence system，而不是 point prediction task。
模型架构/方法流程
- 综述/框架论文，无单一原创预测模型。
- 方法框架包含 health inference、domain shift、cross-fidelity degradation、calibrated uncertainty、BMS-facing validation criteria 的数学化描述。
输入特征与输出目标
- 输入特征：综述覆盖容量、电压/电流/温度、阻抗、热成像、relaxation、数字孪生等多模态证据；具体因原始文献而异。
- 输出目标：SOH、RUL、uncertainty、deployment acceptance criteria、warning risk。
数据集/电池体系
- 电池体系：lithium-ion batteries，覆盖 cell/module/pack/fleet 层级。
- 数据来源/结构：176 篇去重文献 + evidence audit workbook；未能访问全文确认文献筛选细则。
- 跨验证：框架重点正是跨实验室、跨 pack、跨 fleet 的多保真验证。
EOL/RUL/SOH 定义
- SOH/RUL：摘要未给单一定义；强调应针对 deployment criteria、worst-domain error、calibration data 和 warning risk 明确定义。
- EOL 阈值：未能从可访问内容确认统一阈值。
- 标签生成：作为综述框架，重点是审计标签可信度和跨域可迁移性。
验证设计
- 提出的验证阶梯：synthetic cells、controlled aging、module/pack testing、fleet shadow operation、closed-loop deployment。
- 可作为哥哥论文中“外部验证与上线前验收”的框架参考。
指标
- 摘要提到 worst-domain error、calibration data、warning risk、computational feasibility。
- 未能确认具体数值指标表；非单模型实验论文。
Baseline/对比方法
- 横向综述 model-based、ML、DL、transfer learning、physics-informed、impedance、thermographic、relaxation、digital twin。
- 非同一 benchmark 的直接 baseline 比较。
泄漏风险/局限性/复现性
- 泄漏风险：低到中。作为综述不存在模型训练泄漏；但其判断依赖文献审计质量，全文未获取前无法确认审计表细节。
- 局限性：不能提供新模型性能；若 supplementary workbook 不开放，复核成本较高。
- 复现性：信息不足；需获取 MDPI 全文和补充 workbook。
对哥哥研究/落地的启发和后续动作
- 强烈适合做“评价协议”参考：每个 RUL/SOH 模型都应报告最坏域误差、校准、跨协议泛化和计算可行性。
- 后续动作：拿到 PDF 后提取 acceptance criteria，转成哥哥项目的 benchmark checklist。
是否值得跟进
- 结论：强烈跟进（方法论/评价框架用途）。
- 理由：与“可迁移、可信、部署级 SOH/RUL”高度契合，虽然不是原创模型。
查看关键词分类依据
- RUL prediction · 标题包含「rul」
Toward Trustworthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity Vali - SOH estimation · 标题包含「soh」
Toward Trustworthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity - PINN / physics-informed · 摘要包含「physics-informed」
learning, deep learning, transfer learning, physics-informed, impedance-based, thermographic, relaxation-based, and digital twin approaches - Transfer learning / domain adaptation · 摘要包含「transfer learning」
odel-based, machine learning, deep learning, transfer learning, physics-informed, impedance-based, thermographic, relaxation-based, and digita - Generic neural network / deep learning · 摘要包含「deep learning」
w synthesizes model-based, machine learning, deep learning, transfer learning, physics-informed, impedance-based, thermographic, relaxatio - Generic machine learning · 摘要包含「machine learning」
rkbook, this review synthesizes model-based, machine learning, deep learning, transfer learning, physics-informed, impedance-based, thermogra - Bayesian · 摘要包含「uncertainty」
pproaches through observability, robustness, uncertainty calibration, transferability, and deployment feasibility. A compact mathematica - Hybrid model · 摘要包含「closed-loop」
(pack) testing, fleet shadow operation, and closed-loop safety-governed deployment using acceptance criteria, based on worst-domain err - Self-built / experimental dataset · 标题包含「laboratory」
and Multi-Fidelity Validation Framework from Laboratory Cells to Real-World Packs - Li-ion / LIB · 标题包含「lithium-ion」
rthy and Transferable SOH/RUL Estimation for Lithium-Ion Batteries: A Critical Review and Multi-Fidelity Validation Framework from Labor