随着锂离子电池在电动汽车和电力储能领域的大规模应用，频繁发生的电池系统起火爆炸事故严重阻碍了锂离子电池的进一步普及。热失控是锂离子电池安全性事故的根源，电池制造缺陷是导致热失控的潜在原因，其机理复杂且缺少相关研究。因此，本课题聚焦缺陷电池的失效机理、建模与故障诊断，针对金属异物和极耳撕裂这两种可直接影响电池安全性的制造缺陷开展研究。首先，针对电池金属异物缺陷问题，实验揭示了两条不同的金属异物缺陷致内短路的演化路径：电化学路径和机械路径，并论证了金属异物缺陷导致高危害的铝-负极型内短路的可能性。进一步面向电池产线需求，开发了基于局部异常因子的金属异物电池检测方法，并基于实际产线数据对该方法的有效性进行了测试和验证，结果表明其检出率相比于现有方法有较大提升。最后，探索了金属异物电池修复策略，初步实现了内短路故障的修复。然后，针对电池极耳撕裂缺陷问题，实验对比了不同类型极耳撕裂缺陷对电池性能的影响，结果表明负极极耳断裂缺陷是最为危险的缺陷类型，并通过多种表征手段，系统揭示了负极极耳断裂电池的两大失效机理：铜箔溶解和极片边缘析锂。进一步搭建了负极极耳断裂电池的二维电化学模型，模型仿真结合实验验证表明负极极耳断裂电池存在着与正常电池不同的多工况析锂特性，不仅在大倍率充电时会发生大面析锂，其在小倍率充电时也容易发生边缘析锂，最终影响电池安全性。最后，针对制造缺陷和电池滥用导致的共性故障：电池内短路和析锂，开展了电池故障诊断方法的研究。针对实际使用过程中突然死亡型热失控难以检测的问题，提出了面向月级别长周期运行数据的电池微内短路检测方法，为突然死亡型热失控的安全预警提供了新思路。针对传统的基于弛豫曲线的析锂检测方法数据难以获取的问题，本课题提出了适用于动态放电工况的析锂检测方法和基于交流阻抗成分分析的析锂检测方法，并进行了测试验证。所提出的方法有效拓展了现有故障诊断方法的应用场景，为内短路和析锂检测问题提供了实用的解决方案。

With the large-scale application of lithium-ion batteries in electric vehicles and energy storage systems, frequent battery fire and explosion accidents have seriously hindered the further popularity of lithium-ion batteries. Thermal runaway is the root cause of lithium-ion battery safety accidents. Battery manufacturing defects are potential causes of battery thermal runaway, whose mechanism is complex and lack of research. Therefore, this study focuses on the mechanism, modeling and fault diagnosis of battery failure caused by manufacturing defects, targeting two types of manufacturing defects that can directly affect battery safety: metallic foreign matter defect and tab tearing defect.Firstly, the impact of metallic foreign matter defect on battery safety is investigated. Experiments reveal two different evolutionary paths of internal short circuit caused by metallic foreign matter defect: electrochemical path and mechanical path. The possibility that metallic foreign matter defect leads to the highly hazardous aluminum-anode type internal short circuit is also demonstrated. Based on the actual needs of battery production line, a detection method of metallic foreign matter defect based on local outlier factor is developed. The effectiveness of the method is validated using battery production line data. The results show that its detection ratio is greatly improved compared with existing methods. Finally, the restoration strategies of defective battery are explored and reliable repair of internal short circuit is realized. Then, the impact of tab tearing defect on battery safety is investigated. The impact of different types of tab tearing defect on battery performance is experimentally compared. It is found that complete tearing of anode tab is the most dangerous type of tab tearing defect. Through detailed characterization of the defective battery, two major failure mechanisms are revealed: copper foil dissolution and lithium plating at anode edge. Furthermore, a two-dimensional electrochemical model of the defective battery is constructed. Model simulation combined with experimental validation results show that the defective battery is not only susceptible to large-surface lithium plating at large charging current, but also more susceptible to anode-edge lithium plating at low charging current.Finally, battery fault diagnosis methods are developed for the common faults caused by manufacturing defects and battery abuse: internal short circuit and lithium plating. To address the problem that it is difficult to detect the sudden death type thermal runaway during actual use, an internal short circuit detection method for month-level operation data is proposed, which provides a new approach to the early warning of sudden death type thermal runaway. To address the problem that it is difficult to obtain data of lithium plating detection methods based on voltage relaxation curves, this study proposes a lithium plating detection method applicable to dynamic discharging conditions and a lithium plating detection method based on impedance component analysis, both of which are validated through experiments. The proposed methods effectively expand the application scenarios of fault diagnosis methods and provide practical solutions for internal short circuit and lithium plating detection problems.