摘要
提出一种基于长短期记忆(LSTM)网络的锂离子电池荷电状态(SOC)预测方法。以锂离子电池的充放电电流和电压作为模型输入,对锂离子电池SOC进行预测,结果表明LSTM网络的预测精度高于BP神经网络、BP-PSO混合模型和小波神经网络。利用非参数核密度估计方法来计算锂离子电池SOC预测的置信区间,结果表明能够准确计算不同置信水平下锂离子电池SOC预测的不确定性。
A forecast method of the state of charge(SOC)of lithium-ion battery based on the long-short term memory(LSTM)network was proposed.The SOC of lithium-ion battery was forecasted by taking the charge-discharge current and voltage of lithium-ion battery as model inputs.The results show that the forecasting accuracy of LSTMnetwork is higher than that of BP neural network,BP-PSO hybrid model and wavelet neural network(WNN).The nonparametric kernel density estimation(NPKDE)method was used to calculate the confidence interval of the SOC forecast of lithium-ion batteries.The calculation results show that the confidence interval based on NPKDE can accurately calculate the uncertainty of the SOC forecast of lithium-ion batteries at different confidence levels.
作者
李文启
高东学
李朝晖
饶宇飞
顾波
LI Wenqi;GAO Dongxue;LI Zhaohui;RAO Yufei;GU Bo(State Grid Henan Electric Power Company,Zhengzhou 450000,China;State Grid Henan Electric Power Research Institute,Zhengzhou 450052,China;North China University of Water Resources and Electric Power,Zhengzhou 450045,China)
出处
《电器与能效管理技术》
2020年第5期44-50,共7页
Electrical & Energy Management Technology
基金
国网河南省电力公司科技项目(5217021600A3)。
关键词
锂离子电池
荷电状态
长短期记忆网络
预测不确定性
置信区间
lithium-ion battery
state of charge(SOC)
long-short term memory network
forecast uncertainty
confidence interval
