www.ChinaAET.comMeasurementControlTechnologyandInstruments测控技术与仪器仪表基于改进EO-BP神经网络的高压线损预测*徐利美1,闫磊1,李远1,杨射2,任密蜂3(1.国网山西省电力公司,山西太原030021;2.国网山西超高压变电公司,山西太原030021;3.太原理工大学电气与动力工程学院,山西太原030024)摘要:针对高压线损预测精度不高的问题,提出一种基于均衡优化器(EquilibriumOptimizer,EO)和BP神经网络相结合的线损预测模型。首先,为了提高EO算法的寻优能力,利用多种混沌映射关系初始化种群,使种群多样性增加,全局搜索能力得到改善;同时,采用物竞天择概率跳脱策略改进EO算法,使模型依概率跳出局部最优而收敛于全局最优解。其次,采用改进的EO算法对BP神经网络的权值和偏置进行优化,进而改善BP神经网络的预测效果。最后,实验结果证明,所提线损预测模型相对于回归模型、BP神经网络模型、模拟退火算法优化BP神经网络模型和EO优化BP神经网络模型具有更高的预测精度。关键词:线损预测;混沌映射;物竞天择概率跳脱策略;均衡优化器算法;神经网络中图分类号:TP183;TM73文献标志码:ADOI:10.16157/j.issn.0258-7998.223399中文引用格式:徐利美,闫磊,李远,等.基于改进EO-BP神经网络的高压线损预测[J].电子技术应用,2023,49(3):82-88.英文引用格式:XuLimei,YanLei,LiYuan,etal.High-voltagelinelosspredictionbasedonimprovedEO-BPneuralnetwork[J].ApplicationofElectronicTechnique,2023,49(3):82-88.High-voltagelinelosspredictionbasedonimprovedEO-BPneuralnetworkXuLimei1,YanLei1,LiYuan1,YangShe2,RenMifeng3(1.StateGridShanxiElectricPowerCompany,Taiyuan030021,China;2.ShanxiExtraHighVoltageSubstationCompanyofStateGrid,Taiyuan030021,China;3.CollegeofElectricalandPowerEngineering,TaiyuanUniversityofTechnology,Taiyuan030024,China)Abstract:Aimingattheproblemoflowaccuracyofhighvoltagelinelossprediction,alinelosspredictionmodelisproposedbasedonimprovedBPneuralnetworkandEqualizationoptimizer(EO)algorithm.Firstly,inordertoimprovetheoptimizationabilityofEOalgorithm,avarietyofchaoticmappingrelationsisusedtoinitializethepopulationtoincreasethepopulationdiver‐sity,thentheglobalsearchabilitycouldbeimproved.Atthesametime,theEOalgorithmisimprovedby...