灵敏度(控制系统)
计算机科学
电池(电)
启发式
不确定度量化
钥匙(锁)
领域(数学)
可靠性工程
启发式
实验设计
加速寿命试验
超定系统
实验数据
基于模型的设计
数学优化
模拟
压力测试(软件)
忠诚
工作(物理)
持续时间(音乐)
优化设计
数学模型
作者
Florian Stroebl,Florian Schaeufl,Oliver S. Bohlen,Herbert Palm
标识
DOI:10.1016/j.est.2025.119552
摘要
Lithium-ion battery degradation presents challenges for aging characterization due to various stress factors and lengthy testing periods. Optimal experimental design (OED) techniques can enhance study efficiency by maximizing information gain with minimum number of experiments. This study evaluates the efficiency of parameter-individual OED (pi-OED) in calibrating a lithium-ion battery aging model using a multi-stage framework. Analyzing a dataset from 279 commercial cells tested under 71 different conditions, parameter-individual OED is compared with traditional approaches for calendar and cycle aging. For calendar aging, pi-OED reduces uncertainty in the temperature-dependent model parameters by 28% compared to a non-optimized experimental design, demonstrating significant efficiency gains. However, when determining cyclic aging parameters, pi-OED cannot reduce the parameter uncertainty of the underlying heuristic model. This can be explained by its inadequacy, which predicts a model sensitivity that could not be confirmed by experiments. This highlights the necessity of a valid model for successful OED applications. We provide a structured framework for OED in the field of battery aging, addressing the key challenges in terms of effectiveness and efficiency. Our work guides efficient battery characterization experiments and underscores the limitations of model-based optimization when model fidelity or parameter significance are insufficient. • A structured Optimal Experimental Design (OED) framework is proposed for battery studies. • Parameter individual OED cuts calendar aging parameter uncertainty by 28%. • Parameter individual OED shows no benefit for overdetermined models. • Model inadequacy explains failure of parameter individual-OED in cycle aging. • Valid model sensitivity is key for OED to boost efficiency.
科研通智能强力驱动
Strongly Powered by AbleSci AI