Investigation of the Storage Life of Lithium-ion Battery Based on the Metabolism GM(1, 1)-Neural Network
-
摘要: 现役鱼雷大多时间处于贮存状态, 而锂离子电池是其主要的动力能源, 所以针对锂离子电池的贮存寿命研究尤为重要。文中以18650型钴酸锂电池为研究对象, 通过开展加速寿命试验, 获得在不同应力条件下电池容量和内阻随时间的变化曲线, 并确定有利于缓解电池寿命衰减的贮存条件为: 温度25℃、电池荷电状态30%; 综合灰色预测方法及BP神经网络的优点, 采用新陈代谢灰色模型GM(1, 1)-神经网络方法对锂离子电池的容量进行预测, 经验证该组合预测模型比灰色预测模型和新陈代谢GM(1, 1)预测模型精度更高且更适用于电池寿命预测, 从而获取锂离子电池在不同应力条件下的贮存寿命, 进一步验证了有利于缓解电池寿命的贮存条件。Abstract: Active torpedoes are in storage most of the time, and lithium-ion batteries are the main power sources. Therefore, it is important to study the storage life of lithium-ion batteries. In this study, a 18650 lithium cobalt oxide battery was used as the research object. Through the accelerated life test, the battery capacity and internal resistance change curves with time under different stress conditions are obtained, and the storage condition that is helpful to alleviate the attenuation of battery life is determined as 25℃ and 30% state of charge(SOC). This study integrates the advantages of the gray prediction method and neural network, and uses the metabolic GM(1, 1)-neural network method to predict the capacity of lithium-ion batteries. The combined prediction model was verified to be better than the gray prediction model and metabolic GM(1, 1). The prediction model has a higher accuracy and is more suitable for predicting the storage life of lithium-ion batteries under different stress conditions. This further verifies the storage conditions that are conducive to alleviating the attenuation of battery life.
-
Key words:
- lithium-ion battery /
- accelerated life test /
- grey neural network /
- storage life
-
图 1 不同SOC和贮存温度下电池容量变化曲线
Figure 1. Capacity curves of battery under different SOC and storage temperatures
图 2 不同SOC和贮存温度下电池内阻变化曲线
Figure 2. Internal resistance curves of battery under different SOC and storage temperatures
表 1 18650型锂离子电池活化制度表
Table 1. Activation system table of 18650 type lithium ion battery
步骤 具体操作 1 恒流1/3 C放电至2.75 V 2 静置10 min 3 恒流1/3 C充电至4.20 V 4 恒压4.2 V充电至电流下降到0.02 C（51 mA） 5 静置10 min 6 恒流1/3 C放电至2.75 V 
下载: 导出CSV
表 2 GM(1, 1)模型预测结果
Table 2. Prediction results of GM(1, 1) model
时间/d 实际容量/mAh 预测容量/mAh 相对误差/% 0 2568.10 − − 7 2 516.75 2 521.89 0.204 14 2 470.85 2 490.47 0.794 21 2 460.30 2 459.45 0.035 28 2 422.20 2 428.81 0.273 35 2 423.15 2 398.55 1.015 42 2 432.00 2 368.67 2.604 49 2 327.85 2 339.16 0.486 56 2 286.20 2 310.02 1.042 63 2 259.50 2 281.25 0.963 70 2 243.30 2 252.83 0.425 77 2 214.10 2 224.76 0.481 84 2 217.00 2 197.05 0.900
下载: 导出CSV
表 3 新陈代谢GM(1, 1)模型预测结果
Table 3. Prediction results of metabolic GM(1, 1) model
时间/d 实际容量/mAh 预测容量/mAh 相对误差/% 0 2568.10 − − 7 2516.75 2506.15 0.421 14 2470.85 2464.79 0.245 21 2460.30 2464.10 0.154 28 2422.20 2451.64 1.215 35 2423.15 2441.17 0.744 42 2432.00 2400.43 1.298 49 2327.85 2320.65 0.309 56 2286.20 2280.99 0.228 63 2259.50 2262.35 0.126 70 2243.30 2243.86 0.025 77 2214.10 2225.52 0.516 84 2217.00 2207.32 0.437
下载: 导出CSV
表 4 新陈代谢GM(1, 1)-神经网络模型输入与输出样本
Table 4. Input and output samples of metabolic GM(1, 1)-neural network model
序列 输入样本 输出样本 1 $ {x^{(0)}}(1) $ $ {x^{(0)}}(2) $ ··· $ {x^{(0)}}(p) $ $ {e^{(0)}}(p + 1) $ 2 $ {x^{(0)}}(2) $ $ {x^{(0)}}(3) $ ··· $ {x^{(0)}}(p + 1) $ $ {e^{(0)}}(p + 2) $ ··· ··· ··· ··· ··· ··· q−1 $ {x^{(0)}}(q - 1) $ $ {x^{(0)}}(q) $ ··· $ {x^{(0)}}(n - 1) $ $ {e^{(0)}}(n) $ q $ {x^{(0)}}(q) $ $ {x^{(0)}}(q + 1) $ ··· $ {x^{(0)}}(n) $ $ {e^{(0)}}(n + 1) $
下载: 导出CSV
表 5 新陈代谢GM(1, 1)-神经网络模型输入与输出样本值
Table 5. Input and output sample values of metabolic GM(1, 1)-neural network model
序列 输入样本 输出样本 1 2 568.10 2506.15 2446.79 3.80 2 2506.15 2446.79 2446.10 29.44 3 2446.79 2446.10 2451.64 18.02 4 2446.10 2451.64 2441.17 −31.57 5 2451.64 2441.17 2400.43 −7.20 6 2441.17 2400.43 2320.65 −5.21 7 2400.43 2320.65 2280.99 2.85 8 2320.65 2280.99 2262.35 0.56 9 2280.99 2262.35 2243.86 11.42 10 2262.35 2243.86 2225.52 −9.68 11 2243.86 2225.52 2207.32 $ {e^(}^{0)}(1{\text{4}}) $
下载: 导出CSV
表 6 新陈代谢GM(1, 1)-神经网络模型预测结果
Table 6. Prediction results of metabolic GM(1, 1)-neural network model
序列 实际容量/mAh 新陈代谢GM(1, 1)-神经
网络模型预测值/mAh相对误差/% 4 2460.30 2459.60 0.028 5 2422.20 2421.44 0.031 6 2423.15 2422.56 0.024 7 2432.00 2431.69 0.013 8 2327.85 2327.89 0.002 9 2286.20 2286.02 0.008 10 2259.50 2258.99 0.023 11 2243.30 2243.02 0.012 12 2214.10 2210.80 0.149 13 2217.00 2199.31 0.798
下载: 导出CSV
表 7 组合预测模型得出的电池寿命
Table 7. Battery life based on combined prediction model
应力条件 寿命/d 应力条件 寿命/d 10℃、30%SOC 604 40℃、30%SOC 455 10℃、100%SOC 557 40℃、100%SOC 417 10℃、65%SOC 543 40℃、65%SOC 391 25℃、30%SOC 518 55℃、30%SOC 285 25℃、100%SOC 497 55℃、100%SOC 250 25℃、65%SOC 482 55℃、65%SOC 160
下载: 导出CSV
-
[1] 赵涛, 刘明雍, 周良荣. 自主水下航行器的研究现状与挑战[J]. 火力与指挥控制, 2010, 35(6): 1-6.Zhao Tao, Liu Ming-yong, Zhou Liang-rong. A Survey of Autonomous Underwater Vehicle Recent Advances and Future Challenges[J]. Fire Control & Command Control, 2010, 35(6): 1-6. [2] 颜佳辉. 基于多元性能退化的电连接器贮存可靠性建模与验证[D]. 杭州: 浙江理工大学, 2019. [3] 韩鹏飞. 锂电池小电流放电的可靠性分析[J]. 电源技术, 1994(2): 34-37.Han Peng-fei. Analysis on Low-rate Discharge Reliability for Lithium Batteries[J]. Chinese Journal of Power Sources, 1994(2): 34-37. [4] 刘勇. LiFePO_4锂离子电池满电荷存储寿命的研究[D]. 长沙: 国防科学技术大学, 2013. [5] Dubarry M, Truchot C, Liaw B Y. Synthesize Battery Degradation Modes via a Diagnostic and Prognostic Model[J]. Journal of Power Sources, 2012, 219(12): 204-216. [6] 黄燕, 宋保维, 谢亚丽. 基于性能退化数据的锂离子电池贮存可靠性评估方法[J]. 鱼雷技术, 2011, 19(1): 43-47. doi: 10.3969/j.issn.1673-1948.2011.01.010Huang Yan, Song Bao-wei, Xie Ya-li. A Storage Reliability Evaluation Method of Lithium Battery Based on Capability Degenerate Data[J]. Torpedo Technology, 2011, 19(1): 43-47. doi: 10.3969/j.issn.1673-1948.2011.01.010 [7] Thomas W, Marcel W, et al. Temperature Dependent Ageing Mechanisms in Lithium-ion Batteries-A Post-Mortem Study[J]. Journal of Power Sources, 2014, 262: 129-135. [8] Kassem M, Bernard J, Revel R, et al. Calendar Aging of a Graphite/LiFE PO4 Cell[J]. Journal of Power Sources, 2012, 208: 296-305. doi: 10.1016/j.jpowsour.2012.02.068 [9] 王亚辉, 李晓钢. 基于退化轨迹的锂离子电池加速试验研究[J]. 装备环境工程, 2013(2): 30-33.Wang Ya-hui, Li Xiao-gang. Study of Li-ion Cells Accelerated Test Based on Degradation Path[J]. Equipment Environmental Engineering, 2013(2): 30-33. [10] Guan T, Sun S, Gao Y, et al. The Effect of Elevated Temperature on the Accelerated Aging of LiCoO2/Mesocarbon Microbeads Batteries[J]. Applied Energy, 2016, 177: 1-10. [11] 倪江锋, 周恒辉, 陈继涛, 等. 锂离子电池中固体电解质界面膜(SEI)研究进展[J]. 化学进展, 2004, 16(3): 335-342. doi: 10.3321/j.issn:1005-281X.2004.03.003Ni Jiang-feng, Zhou Heng-hui, Chen Ji-tao, et al. Progress in Solid Electrolyte Interface in Lithium Ion Batteries[J]. Progress in Chemistry, 2004, 16(3): 335-342. doi: 10.3321/j.issn:1005-281X.2004.03.003 [12] 闻人红雁, 毛松科, 闻人红权, 等. 不同荷电态对锂离子电池循环寿命的影响[J]. 材料科学与工程学报, 2011, 29(5): 797-799, 769.Wenren Hong-yan, Mao Song-ke, Wenren Hong-quan, et al. Effect of State-of-Charge on the Cycle Life of Lithium-ion Batteries[J]. Journal of Materials Science and Engineering, 2011, 29(5): 797-799, 769. -
点击查看大图
计量
- 文章访问数: 608
- HTML全文浏览量: 22
- PDF下载量: 9
- 被引次数: 0
