An indirect remaining useful life prognosis for Li-ion batteries based on health indicator and novel artificial neural network
An indirect remaining useful life prognosis for Li-ion batteries based on health indicator and novel artificial neural network
复制标题
基于健康指标和新型人工神经网络的锂离子电池剩余使用寿命间接预测
DOI：
10.1016/j.est.2022.104701
复制
发表时间：
2022-08
影响因子：
9.4
通讯作者：
Huimei Yuan
中科院分区：
文献类型：
--
作者：
Ting Tang;Huimei Yuan
准确预测电池剩余使用寿命（RUL）对电池的安全性、可靠性和经济性具有重要意义。本文在充电和放电过程中，对所有测量到的电压、电流、温度等参数进行大范围搜索。针对参数的不同特征，提出了简单而又不同的特征提取方法，以尽可能获得与容量相关的健康指标。此外，根据异常点的位置，采用两种校正方法对提取的HIs中可能存在的异常点进行改进。一种是均值代换法，另一种是标准化代换法。然后，根据每个HI与容量之间的Pearson相关系数进行最优特征选择，选择系数值最高的HI作为最优HI。最后，为了建立HI与容量的关系，从时间序列HI的形状特征、时变特征和序列变换特征三个方面构建人工神经网络（ANN），实现RUL预测。在不同的实验条件下，通过NASA和Oxford的数据集验证了该方法的有效性和泛化性。
Accurate prediction of the remaining useful life (RUL) plays an important role in safety, reliability and economy for batteries. In this paper, all the measured voltage, current, temperature and other parameters are searched in a wide range during charging and discharge process. Based on the different characteristics of parameters, simple but different feature extraction approaches are proposed to obtain the health indicators (HIs) related to capacity as far as possible. In addition, according to the location of outliers, two correction approaches are used to improve possible outliers in the extracted HIs. One is the mean substitution approach, and the other is the normalized substitution approach. Then, based on the Pearson correlation coefficient between each HI and capacity, the optimal feature selection is performed, and the HI with the highest coefficient value is selected as the optimal HI. Finally, in order to establish the relationship between HI and capacity, an artificial neural network (ANN) is constructed from three aspects of the shape feature, time-dependent feature and sequence transformation feature of time series HI, so as to realize RUL prediction. The effectiveness and generalization of the proposed approach are verified by data sets from NASA and Oxford under different experimental conditions.
登录
查看更多内容
使用非生命周期标记数据集进行 RUL 预测的平均 Bi-LSTM 网络
影响因子：
6
作者：
Yu Yong;Hu Changhua;Si Xiaosheng;Zheng Jianfei;Zhang Jianxun
通讯作者：
Zhang Jianxun
影响因子：
9.4
作者：
Lyu Zhiqiang;Geng-feng Wang;Renjing Gao
通讯作者：
Lyu Zhiqiang;Geng-feng Wang;Renjing Gao
影响因子：
9.2
作者：
Yapeng Zhou;Miaohua Huang;Yupu Chen;Ye Tao
通讯作者：
Yapeng Zhou;Miaohua Huang;Yupu Chen;Ye Tao
影响因子：
1.6
作者：
Su, Xiaohong;Wang, Shuai;Ye, Zhe
通讯作者：
Ye, Zhe
影响因子：
11.2
作者：
Haris, Muhammad;Hasan, Muhammad Noman;Qin, Shiyin
通讯作者：
Qin, Shiyin