摘要
储能电站电池管理系统作为实时、连续监测电池的重要二次设备,缺乏合理的健康度评估与预警方法。为此,提出一种基于最优径向基函数核主成分和双向长短期记忆网络结合的健康度评估与预警方法。传统核主成分在储能电站电池管理系统监测数据特征提取时存在核参数选择困难的问题,通过交叉验证方法选取最优核参数,结合K邻域多维尺度分析方法实现样本重构,提升系统对健康状态影响因素进行特征提取的能力;通过构建双向长短期记忆网络模型以输入特征信息达到健康预警目的。以某储能电站采集到的真实数据作为样本,通过试验数据进行对比分析。结果表明,所提方法可有效提升电池管理系统中多维健康监测数据的准确评估和预测精度,为检修人员制定检修策略提供科学参考。
The battery management system(BMS)of energy storage power station,as an important secondary device for real-time and continuous monitoring of batteries,lacks reasonable health assessment and warning methods.Therefore,a health assessment and warning method based on the combination of optimal radial basis function kernel principal component analysis(PCA)and bidirectional long short-term memory(LSTM)network was proposed.Traditional kernel PCA had the problem of difficulty in selecting kernel parameters in battery management system monitoring data feature extraction,the cross-validation method was,therefore,used to select the optimal kernel parameters,combined with K-neighborhood multidimensional scale analysis method to achieve sample reconstruction and improve the system's ability to extract features of factors affecting health status;by constructing a bidirectional LSTM network model to input feature information,the goal of health warning was achieved.The real data collected from a certain energy storage power station was used as a sample for comparative analysis through experimental data.The results indicate that the proposed method can effectively improve the accurate evaluation and prediction accuracy of multidimensional health monitoring data in battery management system,providing scientific reference for maintenance personnel to formulate maintenance strategies.
作者
肖峰
刘新超
杨欢红
朱伟星
周春峰
叶婧元
Xiao Feng;Liu Xinchao;Yang Huanhong;Zhu Weixing;Zhou Chunfeng;Ye Jingyuan(Shanghai Huadian Fengxian Thermal Power Co.,Ltd.,Shanghai 201499,China;Shanghai University of Electric Power,Shanghai 200090,China)
出处
《电气自动化》
2025年第3期64-69,共6页
Electrical Automation
基金
中国华电集团公司科技项目(CHDKJ24-04-02-59)。
关键词
电池管理系统
核主成分
双向长短期记忆网络
健康度评估
健康度预警
battery management system
kernel principal component
bidirectional long short-term memory network
health assessment
health early warning
