| 引用本文: | 梁海峰,袁芃,高亚静.基于CNN-Bi-LSTM网络的锂离子电池剩余使用寿命预测[J].电力自动化设备,2021,41(10): |
| LIANG Haifeng,YUAN Peng,GAO Yajing.Remaining useful life prediction of lithium-ion battery based on CNN-Bi-LSTM network[J].Electric Power Automation Equipment,2021,41(10): |
|
| 摘要: |
| 锂离子电池的剩余使用寿命(RUL)预测可以评估电池的可靠性，降低电池使用的风险并为电池维护提供理论依据。结合卷积神经网络(CNN)与双向长短期记忆(Bi-LSTM)网络的优点，提出一种考虑多种寿命衰退特征与数据时序性的CNN-Bi-LSTM网络模型用于锂离子电池RUL预测。通过仿真得到CNN超参数，选择相关性高的特征参数作为预测输入量，最后在 NASA 锂离子电池老化数据集上进行仿真实验。实验结果表明CNN-Bi-LSTM网络模型能准确预测锂离子电池RUL，与其他网络模型相比，具有网络模型参数少、占用内存小的优势，在精确度和收敛性上都有较好表现。 |
| 关键词: 锂离子电池 卷积神经网络 双向长短期记忆网络 剩余使用寿命预测 |
| DOI：10.16081/j.epae.202110030 |
| 分类号:TM911 |
| 基金项目:国家自然科学基金资助项目(51607068) |
|
| Remaining useful life prediction of lithium-ion battery based on CNN-Bi-LSTM network |
|
LIANG Haifeng1, YUAN Peng1, GAO Yajing2
|
|
1.Department of Electrical Engineering, North China Electric Power University, Baoding 071003, China;2.Carbon Neutrality Research Institute of China Huaneng Group Co.,Ltd.,Beijing 100031, China
|
| Abstract: |
| The RUL(Remaining Useful Life) prediction of the lithium-ion battery can evaluate the reliability of the battery, reduce the risk of battery use and provide a theoretical basis for battery maintenance. Combining the advantages of CNN(Convolutional Neural Network) and Bi-LSTM(Bi-directional Long Short-Term Memory) network, the CNN-Bi-LSTM network model for lithium-ion battery RUL prediction is proposed, which considers both multiple degradation characteristics and time sequence. The hyperparameters of CNN are obtained by simulation, the highly correlated feature parameters are selected as the prediction input, and the simulation experiment is carried out on the NASA lithium-ion battery aging data set. The experimental results show that the CNN-Bi-LSTM network model can accurately predict the RUL of lithium-ion batteries. Compared with other network models, it has the advantages of fewer network model parameters and smaller memory usage, and has good performance in accuracy and convergence. |
| Key words: lithium-ion battery convolutional neural network bi-directional long short-term memory network remaining useful life prediction |
