摘要
针对现有研究中的不足,提出一种改进的相关向量机模型对锂离子电池的剩余寿命进行实时预测。首先,利用锂离子电池容量数据和相空间重构技术,构造模型的训练集。其次,在传统的单核相关向量机模型上做出改进,采用多核核函数,从而提高相关向量机模型的泛化能力和剩余寿命预测的精度。此外,利用粒子群优化算法,自适应的确定多核相关向量机模型的最优参数组合。实验结果表明,相比单核相关向量机模型,本文提出的多核相关向量机模型能够更为准确的对锂离子电池的剩余寿命进行预测。
Considering the existing research gap,we propose a modified Relevance Vector Machine（ RVM） to make an online prediction of RUL. Firstly,the training dataset with lithium-ion batteries＇ capacity data and phase space reconstruction technique is constructed. Then,the traditional RVM by adopting a multi-kernel function is modified in order to improve the accuracy and generalizability of the original RVM model. In addition,the particle swarm optimization algorithm is employed to adaptively search the optimal parameter combination of the multi-kernel RVM. The experimental results show that the present model is capable to predict the online RUL of lithium-ion battery more accurate than that of single kernel RVM.
出处
《机械科学与技术》
CSCD
北大核心
2016年第8期1286-1290,共5页
Mechanical Science and Technology for Aerospace Engineering
基金
国家自然科学基金面上项目(71371182)资助
关键词
锂离子电池
多核相关向量机
粒子群优化
相空间重构
剩余寿命
lithium-ion battery
multi-kernel relevance vector machine
particle swarm optimization
phase space reconstruction
remaining useful lifetime
