资源描述:
《基于ARM的光伏电站数据监测系统设计.doc》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、第44卷第504期电测与仪表Vol.44No.5042007年第12期ElectricalMeasurement&InstrumentationDec.2007基于卡尔曼滤波预测的电动汽车分时租赁监测数据去人为干预技术研究*龙羿,侯兴哲,肖剑锋,孙洪亮,刘永相,朱彬(国网重庆市电力公司电力科学研究院,重庆401123)摘要:随着社会环保意识的增强,大力推广采用清洁能源的电动汽车,电动汽车分时租赁业务随之不断拓展,多级管理平台纷纷建立。总管理平台所采集的电动汽车参数日益积累,为大数据分析研究奠定了坚实的基础。预处理技术在数据采集过程中起着至关重要的作用。特别地,从各运营商子
2、平台所采集的监测数据存在人为干预风险。为此,文章提出采用人为干预概率曲线量化监测数据与协方差比间关系,将人为干预概率曲线区间和观测量关系作为输入,建立了回归对象决策树并引入传统卡尔曼滤波算法,从而提出基于决策树分析的卡尔曼滤波预测方法,从而减少人为篡改的干预影响,以达到对运营总平台所采集的数据去人为干预的目的。将该算法应用于车速预测领域,得到可信度更高的车速预测数据,为有效地实现大数据分析奠定坚实的技术支撑。关键词:电动汽车;分时租赁;车速预测;卡尔曼滤波算法;人为干预中图分类号:TM863文献标识码:B文章编号:1001-1390(2017)21-0000-00Thes
3、Studyofeliminatinghumaninterventionforelectricvehicletime-sharingleasingmonitoringdatabasedonamodifiedKalmanfilterpredictionalgorithmLongYi,HouXingze,XiaoJianfeng,SunHongliang,LiuYongxiang,ZhuBin(ElectricPowerResearchInstituteofStateGridChongqingElectrictPowerCompanyo,ElectPowerResInst,Ch
4、ongqing401123,China)Abstract:Withtheenhancementofsocialenvironmentalprotectionconsciousness,electricvehiclesusingcleanenergyarepromotedwidely.TheScopescopeoftime-sharingleasingbusinessfortheelectricvehiclekeepsexpanding,andmulti-levelmanagementplatformsaresetuponeafteranother.Electricvehi
5、cleparameterscollectedbymainmanagementplatformareaccumulatedcontinually,whichlaysafirmfoundationforbigdataanalysisresearch.Preprocessingtechnologyplaysanimportantroleinontheprocessofdataacquisition.EsSpecially,monitoringdatacollectedfromtheeachsub-platformofoperatorsmayhavetheriskofbeingt
6、amperedmanually.Therefore,inthispaper,humaninterventionprobabilitycurveisusedtoquantifytherelationshipbetweenmonitoringdataandcovarianceratio,andbothofhumaninterventionprobabilitycurveintervalandobservationdatarelationshiparetreatedasinput,sothattheregressionobjectdecisiontreeisestablishe
7、dandisintroducedtothetraditionalKalmanfilteralgorithm.Therefore,amodifiedKalmanfilterpredictionmethodbasedondecisiontreeanalysisispresentedwhichcanbeusedtoreducetheinfluenceofhumantampering,andachievepreprocessingfunctionofmainoperatingplatformforcollectingdata.Inth