小样本下基于改进麻雀算法优化卷积神经网络的飞轮储能系统损耗.pdfVIP

49卷1期电网技术Vol.49No.1

2025年1月PowerSystemTechnologyJan.2025

文章编号：1000-3673（2025）01-0366-07中图分类号：TM721文献标志码：A学科代码：47040

小样本下基于改进麻雀算法优化卷积神经网络的

飞轮储能系统损耗

魏乐，李承霖，房方，刘渝斌

（华北电力大学控制与计算机工程学院，北京市昌平区102206）

OptimizationofConvolutionalNeuralNetworkBasedonImprovedSparrowAlgorithmfor

FlywheelEnergyStorageSystemLossinSmallSample

WEILe,LIChenglin,FANGFang,LIUYubin

(SchoolofControlandComputerEngineering,NorthChinaElectricPowerUniversity,ChangpingDistrict,Beijing102206,China)

1ABSTRACT:Flywheelenergystoragesystemshavestandby麻雀搜索算法；Logistic混沌映射

lossesandareunsuitableforlong-termenergystorage.Aiming

DOI10.13335/j.1000-3673.pst.2023.2001

attheeconomicindexofflywheelloss,basedonthesmall

sampledataoftheoperationoftheflywheelenergystorage0引言

system,aflywheellosscalculationmodelcombiningaLogistic

风能、太阳能等新能源出力具有间歇性、波动

chaoticsparrowoptimizationalgorithmandaconvolutional

neuralnetworkwasproposed.Firstly,thecausesofflywheel性的特点，新能源的大规模接入将直接影响电力系

losswereanalyzed.Next,theflywheeloperationdataof统可靠性和稳定性。储能系统可用来提升电网电能

NingxiaLingwuPowerPlantwaspreprocessed,andthe质量，飞轮储能系统(flywheelenergystorage

adversarialgenerationnetworkwasusedforsmall-samplesystem，FESS)适用于电力系统调频、间歇式新能源

enrichment.Then,thelossmodelisestablishedbasedonthe

发电、不间断电源等领域，具有充放电迅速、储能

convoluti