摘要
为保证电池的正常运行,对电池的健康状态进行估计是非常重要的工作。针对传统建模方法估计精度差、参数众多计算复杂且耗时长等缺点,本工作构建了基于改进的萤火虫算法(firefly algorithm,FA)优化的反向传播(back propagation,BP)神经网络,对锂离子电池的健康状态(state of health,SOH)进行估计,利用萤火虫算法的全局优化能力和收敛速度快的特点对BP神经网络的权值和阈值进行优化,并引入莱维飞行(Levy flight),提升全局搜索能力,扩大搜索范围,提高了估计精度。采用NASA艾姆斯研究中心的锂离子电池数据集,对改进优化前后的算法进行训练与估计,对比各算法之间的优劣程度。结果表明,莱维飞行改进萤火虫算法优化反向传播神经网络(LFFA-BP)算法相比于BP神经网络算法与萤火虫算法优化反向传播神经网络(FA-BP)算法,决定系数更高,误差波动范围更小,具有较高的估计精度。
It is essential to estimate a battery's state of health.This paper constructs a back propagation(BP)neural network optimized based on the improved firefly algorithm(FA)to estimate the state of health of lithium-ion batteries.The aim is to address the shortcomings of traditional modeling methods,such as poor estimation accuracy,numerous parameters,complex calculation,and longtime consumption.The weights and thresholds of the BP neural network are optimized using the FA's global optimization ability and fast convergence speed.Levy flight is introduced to improve the global search ability,expand the search range,and improve the estimation accuracy.The lithium-ion battery dataset of NASA Ames Research Center is used to train and estimate the algorithms before and after improvement and optimization and compare the advantages and disadvantages of each algorithm.Results show that compared with the BP neural network algorithm and FA for optimizing the BP neural network(FA-BP)algorithm,Levy flight improved firefly algorithm to optimize the BP neural network(LFFA-BP)algorithm has a higher determination coefficient,smaller error fluctuation range,higher estimation accuracy,and specific practical value.
作者
赵鑫浩
许亮
ZHAO Xinhao;XU Liang(College of Electrical Engineering and Automation,Tianjin University of Technology,Tianjin 300382,China)
出处
《储能科学与技术》
CAS
CSCD
北大核心
2023年第3期934-940,共7页
Energy Storage Science and Technology
关键词
锂电池
健康状态
莱维飞行
萤火虫算法
BP神经网络
lithium battery
state of health
Levy flight
firefly algorithm
BP neural network
