基于滑动窗口的锂离子电池单体电压信号在线变分模态分解分析  

Online variational mode decomposition analysis of battery cell voltage signals based on sliding window

作　　者：饶博 梁兆杰 杜进桥 田杰 姜久春 RAO Bo;LIANG Zhaojie;DU Jinqiao;TIAN Jie;JIANG Jiuchun(Shenzhen Power Supply Co.,Ltd.,National Energy Innovation Platform--Source-Grid-Load Interaction R&D Center,Guangdong Provincial Key Laboratory of Source-Grid-Load-Storage Interactive Collaborative Technology,Shenzhen 518000,China;Shenzhen Automotive Research Institute,Beijing Institute of Technology,Shenzhen 518000,China)

机构地区：[1]深圳供电局有限公司国家能源新型电力系统源网荷双向互动研发中心、广东省源网荷储互动协同技术企业重点实验室,广东深圳518000 [2]北京理工大学深圳汽车研究院,广东深圳518000

出　　处：《电气应用》2025年第4期52-57,共6页Electrotechnical Application

基　　金：国家重点研发计划(2022YFB2404300)。

摘　　要：在清洁能源转型的推动下,电池储能系统成为提升电网韧性的重要手段。国际能源署预测,到2030年,全球电池储能装机容量将显著增长,平均持续时间也将延长。在此背景下,电池储能系统的安全管理研究尤为重要。单体电池电压作为关键参数,其信号具有非线性和非平稳特性,异常信号容易因电池老化而难以辨识,因此需要深入的时频域分析。评估了常用的非平稳信号分解方法,并指出其实时性和抗噪能力的不足。相较之下,变分模态分解能有效提取潜在故障特征,适用于动态信号分析。为实现边缘计算环境下的实时在线监测,提出一种基于滑动窗口的在线变分模态分解方法,通过捕捉信号的时变特性和提取模态分量,为电池性能评估和故障诊断提供数据支持。Driven by the transition to clean energy,battery energy storage systems have become a crucial means of enhancing grid resilience.The International Energy Agency(IEA)predicts that by 2030,the global installed capacity of battery energy storage will grow significantly,with average discharge durations also increasing.Against this backdrop,research on the safety management of BESS has become critically important.As a key parameter,single-cell battery voltage exhibits nonlinear and non-stationary characteristics,making it difficult to identify abnormal signals masked by battery aging,thus necessitating in-depth time-frequency domain analysis.This paper evaluates commonly used methods for non-stationary signal decomposition and highlights their limitations in realtime performance and noise resistance.In contrast,Variational Mode Decomposition(VMD)effectively extracts potential fault features,making it suitable for dynamic signal analysis.To achieve real-time online monitoring in edge computing environments,this paper proposes an online VMD method based on a sliding window.By capturing the time-varying characteristics of signals and extracting modal components,the method provides data support for battery performance evaluation and fault diagnosis.

关 键 词：变分模态分解(VMD) 电池单体电压 滑动窗口 在线 

分 类 号：TM912[电气工程—电力电子与电力传动]