适用于全生命周期容量损失机制诊断的锂离子电池OCV老化模型研究

OCV Aging Model of Lithium-ion Batteries for Whole-life Capacity-loss Mechanism Diagnosis

锂离子电池的容量损失是可循环锂损失机制与多种活性材料损失机制共同作用的结果,整体容量损失相同的电池可能由于各机制占比的不同,呈现出截然不同的剩余生命衰减轨迹。如何对电池老化过程中各机制的损失情况进行准确的诊断,对于电池的全生命周期利用至关重要。为此,以更适合于免训诊断的开路电压(Opencircuitvoltage,OCV)老化模型为研究对象,针对其在电池老化后期的表征精度局限,提出面向全生命周期容量损失机制诊断的改进OCV老化模型。新模型充分考虑电池老化后期,活性材料深度损失对电极电势曲线形式的复合影响规律,在现有模型的基础上,通过对其所对应基础电极电势联合坐标系的非均匀压缩扩展,实现对电池OCV全生命老化行为的有效机理描述。为了验证新模型在全生命周期中的有效性,以可用容量衰减至30%的磷酸铁锂电池(Lithium iron phosphate battery,LFP)为试验对象,对其在不同老化程度下开展模型拟合诊断测试。结果表明,在整个生命周期中新模型的RMS误差均被控制在2 mV以内。此外,为了保证新模型在诊断过程中的可靠性,还对其参数可辨识性进行了分析与论证。

The capacity loss of batteries results from the joint effects of the loss mechanisms of recyclable lithium and various active materials.The batteries with the same overall capacity loss may present completely different residual life decay trajectories due to the different proportions of different mechanisms.How to accurately diagnose the loss of each mechanism during battery aging is crucial for batteries’utilization in the whole-life cycle.Therefore,an open circuit voltage(OCV)aging model which is more suitable for training-free diagnosis is selected as the study object,and aiming at the limitation of characterization accuracy at the later stage of battery aging,an improved OCV aging model for capacity-loss mechanism diagnosis in the whole-life cycle is proposed.The new model,fully considering the composite influence law of the deep loss of active material on electrode potential curves at the later stage of battery aging,realizes the effective mechanism description of the whole-life aging behavior of battery OCVs through the non-uniform compression expansion of the corresponding basic electrode potential joint coordinate system,based on the existing model.In order to verify the effectiveness of the new model in the whole-life cycle,lithium iron phosphate batteries(LFP)with available capacity decaying to 30%is taken as the experimental object,and the model fitting diagnostic tests are carried out under different aging states.The results show that the RMS error of the new model is controlled within 2 mV throughout the whole-life cycle.Furthermore,in order to ensure the reliability of the new model in the diagnosis process,its parameter identifiability is also analyzed and demonstrated.