摘要
应用传统的粒子滤波(PF)算法估计锂电池健康状态(SOH)时,会出现粒子权值退化和样本贫化而导致预测精度较低的问题。为了解决该问题,本工作提出了基于改进蝴蝶优化算法和粒子滤波(IBOA-PF)的联合算法,在基本蝴蝶优化算法(BOA)的基础上,用混沌数替代固定的切换概率,并引入共生生物搜索的互生阶段,弥补了蝴蝶算法易陷入局部最优和开发能力差的局限性,提高了BOA的收敛速度,再用蝴蝶表示粒子,用蝴蝶向食物移动的过程表示粒子变化为更符合真实后验分布的新采样值。然后基于双指数经验模型和时间指标(TI)构建了非线性系统的状态空间模型,用单纯形法改进高斯牛顿法进行参数拟合,提出了一种基于IBOA-PF的锂电池SOH估计方法。仿真实验结果表明,该方法优于传统PF方法,具有较高的精确度和较好的适应性。
When the traditional particle filter(PF)algorithm is used to estimate the state of health(SOH)of lithium-ion batteries,several problems arise,such as particle weight degeneration and species decrease,leading to lower prediction accuracy.In this paper,a novel hybrid algorithm,the improved butterfly optimization algorithm based on PF(IBOA-PF),is proposed to solve these problems.This algorithm based on the basic butterfly optimization algorithm(BOA)replaces the stable switching probability with the chaotic maps.It uses the mutualism phase of symbiosis organism search to make up for the limitations of the butterfly algorithm(i.e.,it easily falls into the local optimum and has poor development ability)and improve the convergence speed of BOA.Butterflies are used to represent particles,and the process of butterflies moving to the food is similar to the change of particles having better values that are more possibly equal to the true values.This paper proposed an SOH estimation method using IBOA-PF for lithium batteries based on the double exponential model and time index(TI),constructed the state-space model of the nonlinear system,used the simplex method to improve the Gauss-Newton method for parameter fitting,and estimated SOH.The simulation results show that this method is superior to the traditional PF method,with higher accuracy and better adaptability.
作者
李鹏
李立伟
杨玉新
LI Peng;LI Liwei;YANG Yuxin(School of Electrical Engineering of Qingdao University;Weihai Innovation Institute of Qingdao University;Library of Qingdao University,Qingdao 266071,Shandong,China)
出处
《储能科学与技术》
CAS
CSCD
北大核心
2021年第2期705-713,共9页
Energy Storage Science and Technology
基金
山东省自然科学基金项目(Y2008F23)
山东省科技发展计划项目(2011GGB01123)
山东省重点研发计划项目(2017GGX50114)。
关键词
电池健康状态
粒子滤波
改进蝴蝶优化算法
改进高斯牛顿法
state of health of battery
particle filter
improved butterfly optimization algorithm
improved Gauss-Newton algorithm
