基于Bi-LSTM-Dropout的蓄电池剩余使用寿命预测方法
2024-04-13黄晓智张华明黄艺航许志杰
黄晓智 张华明 黄艺航 许志杰



本文引用格式:黄晓智,张华明,黄艺航,等.基于Bi-LSTM-Dropout的蓄电池剩余使用寿命预测方法[J].自动化与信息工程, 2024,45(1):42-46;60.
HUANG Xiaozhi, ZHANG Huaming, HUANG Yihang, et al. Prediction method of battery remaining useful life based on Bi-LSTM-dropout[J]. Automation & Information Engineering, 2024,45(1):42-46;60.
摘要:蓄电池剩余使用寿命预测对能源的安全性和可持续发展至关重要。该文提出一种蓄电池剩余使用寿命的预测方法,利用蓄电池的历史运行数据和充放电周期,构建Bi-LSTM-Dropout网络模型。利用Bi-LSTM提取时间序列中蓄电池长期依赖的特征,采用Dropout优化算法降低Bi-LSTM网络模型的复杂度,提高模型的泛化能力。实验结果表明,该方法在测试集上的准确率达96.2%,实现了蓄电池剩余使用寿命的精确预测。
关键词:蓄电池;剩余使用寿命预测;Bi-LSTM;Dropout优化算法
中图分类号:TM912 文献标志码:A 文章编号:1674-2605(2024)01-0007-06
DOI:10.3969/j.issn.1674-2605.2024.01.007
Prediction Method of Battery Remaining Useful Life Based on Bi-LSTM-Dropout
HUANG Xiaozhi1 ZHANG Huaming2 HUANG Yihang1 XU Zhijie1
(1.School of Electromechanical Engineering, Guangdong University of Technology, Guangzhou 510006, China
2.School of Advanced Manufacturing Guangdong University of Technology, Jieyang 522000, China)
Abstract: The prediction of the remaining useful life of battery is crucial for the safety and sustainable development of energy. This article proposes a prediction method for the remaining useful life of battery, using historical operating data and charging and discharging cycles of battery to construct a Bi-LSTM-Dropout network model. Using Bi-LSTM to extract long-term dependent features of battery in time series, using Dropout optimization algorithm to reduce the complexity of Bi LSTM network model and improve its generalization ability. The experimental results show that the accuracy of this method on the test set reaches 96.2%, achieving accurate prediction of the remaining useful life of the battery.
Keywords: battery; remaining useful life; Bi-LSTM; dropout optimization algorithm
0 引言
蓄電池组作为变电站直流电源系统的关键设备,在交流电源发生断电时,能为重要负载设备提供不间断的直流供电保障[1]。目前,变电站的通信用蓄电池每年都要进行充放电核容和激活,需要耗费2人3日的工作量[2]。且仅依赖人工使用放电仪表进行核容,无法准确地掌握蓄电池的健康情况[3]。
近年来,国内外学者对蓄电池剩余使用寿命(re-
maining useful life, RUL)预测进行了大量的研究,主要方法包括基于物理模型法和基于数据驱动法[4]。基于物理模型法通过研究电池的内部状态,结合相关算法预测电池RUL[5]。文献[6]利用简化的电化学模型和双非线性滤波器,结合电池的荷电状态(state of charge, SOC)和健康状态(state of health, SOH)估计算法,实现电池RUL的预测。文献[7]基于固体电解质界面副反应建立电化学-热耦合模型,能有效地模拟固体……p>
