Tikhonov正则化
正规化(语言学)
过度拟合
介电谱
电阻抗
数学优化
算法
计算机科学
材料科学
分析化学(期刊)
数学
化学
电化学
电气工程
数学分析
工程类
反问题
人工智能
电极
物理化学
色谱法
人工神经网络
作者
Pengcheng Niu,Kun Yang,Zhengxiang Song,Zheyuan Pang,Zhenping Feng,Jinhao Meng
标识
DOI:10.1016/j.jpowsour.2024.234489
摘要
The distribution of relaxation times (DRT) analysis based on Tikhonov regularization is widely used to analyze the battery electrochemical impedance spectroscopy (EIS), but the selection of the regularization parameter has always been a problem. This paper proposes an Optimization Pre-calculation DRT-tool of EIS (OPDEIS) which not only completes the full-frequency EIS fitting with the insertion capacitor and inductance but also obtains the optimal regularization parameter and the fitting kernel function width through a multi-objective optimization algorithm. In particular, it can resolve the DRT parameter selection for EIS above 0.1 Hz in lithium-ion batteries effectively. This optimizer is suitable for lithium-ion battery EIS data with different states of charge (SOCs) and frequency points. It provides a reference for selecting the regularization parameter and full width at half maximum (FHWM) automatically, enabling the DRT process based on Tikhonov regularization to perfectly avoid overfitting or underfitting.
科研通智能强力驱动
Strongly Powered by AbleSci AI