中北大学学报（自然科学版）2025，Vol.46Issue(3)：316-325,10.DOI:10.62756/jnuc.issn.1673-3193.2024.05.0016
基于N-BEATS和相关向量机的锂电池健康状态混合预测方法
A Hybrid Method Based on N-BEATS and Relevance Vector Machine for Predicting State of Health of Lithium-Ion Batteries
摘要
Abstract
As a core energy component in many fields,accurate prediction of state of health(SOH)is crucial for lithium-ion battery during their lifecycle.A hybrid prediction method based on Neural Basis Expansion Analysis(N-BEATS)and relevance vector machine(RVM)was proposed.Firstly,variational mode decomposition was used to decompose the original time series to improve the accuracy of prediction;Secondly,the decomposed subsequence was divided into high-frequency and low-frequency subsequences based on the center frequency,and the deep neural network N-BEATS model with residual principle and RVM model were used to model and predict them respectively;Finally,the prediction results of each subsequence were overlaid and reconstructed to obtain the final prediction results.To verify the effectiveness of the proposed method,this paper conducted simulation experiments using lithium-ion battery data provided by NASA and CALCE.The experimental results show that compared with the single N-BEATS model and the RVM model,the proposed hybrid method can effectively combine the advantages of the two models and demonstrate higher prediction accuracy.Furthermore,compared with the long short-term memory network,Gaussian process regression,and support vector regression models,the root mean square error of the proposed method is reduced by about 96.5%,74.5%,and 62.5%,and the mean square error is reduced by 97.3%,76.7%,and 58.8%,respectively.关键词
锂离子电池/健康状态/N-BEATS模型/变分模态分解/相关向量机Key words
lithium-ion battery/state of health/N-BEATS model/variational mode decomposition/rel-evance vector machine分类
动力与电气工程引用本文复制引用
李泽龙,乔钢柱,崔方舒,蔡江辉..基于N-BEATS和相关向量机的锂电池健康状态混合预测方法[J].中北大学学报（自然科学版）,2025,46(3):316-325,10.基金项目
山西省基础研究计划资助项目(202303021222084) （202303021222084）
山西省研究生教育创新项目(2024KY613) （2024KY613）
