资源描述：
《EMD与NARX神经网络的风电场总功率组合预测》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、ComputerEngineeringandApplications计算机工程与应用2016，52（12）265EMD与NARX神经网络的风电场总功率组合预测1，2333张振华，马超，徐瑾辉，欧阳泽拯1，2333ZHANGZhenhua,MAChao,XUJinhui,OUYANGZezheng1.广东外语外贸大学经济贸易学院统计系，广州5100062.考文垂大学商务、环境和社会学院，英国考文垂市CV15FB3.广东外语外贸大学金融学院，广州5100061.DepartmentofStatistics,Schoolof
2、EconomicsandTrade,GuangdongUniversityofForeignStudies,Guangzhou510006,China2.FacultyofBusiness,EnvironmentandSociety,CoventryUniversity,CoventryCV15FB,UnitedKingdom3.SchoolofFinance,GuangdongUniversityofForeignStudies,Guangzhou510006,ChinaZHANGZhenhua,MAChao,XUJ
3、inhui,etal.Noveltotal-powercombinationalforecastingmethodofwindfarmbasedonEMDandNARXneuralnetwork.ComputerEngineeringandApplications,2016,52（12）：265-270.Abstract：Ahigh-precisioncombinationalmethodispresentedtoforecastthetotal-powerofwindfarmdirectly.Takingintoac
4、countthattheNon-stationaryidentityofwindspeedleadstothenon-stationarytimeseriesoftotal-power,theNARXneuralnetworkisadoptedastheoriginalforecastingmodel.Andthen,ahybridforecastingmethodbasedonEmpiricalModeDecomposition（EMD）andNARXneuralnetworkisproposedtoimprovet
5、heforecastprecision.First,thetotal-powertimeseriesisdecomposedintoseveralstabletrendtermswithdifferentIntrinsicModeFunctions（IMF）.Subsequently,thecorrespondingpredictionmodelsofNARXneuralnetworkaresetupaccordingtodifferentstablecomponents.Theseforecastingresults
6、ofeachcomponentmodelaresummedupinequalweighttoobtainthefinalpredictivevalue.Besides,thedataoftimeintervalsof5-minuteand15-minuteobtainedfromalargewindpowerplantareusedintheexperimentstoexplorehowtimeintervalsaffectthepredictiveresults.Theexperimentshowsthattheco
7、mbinationalmodelissuitableforthepredictionoftotal-power.Accordingtotheexperimentalresults,thecombinationalmodelismoreaccuratethanmanytraditionalmethods,andthepredictionaccuracyof5-minutetimeintervaldataismoreaccuratethanthatof15-minutetimeintervaldata.Keywords：e
8、mpiricalmodedecomposition;NonlinearAuto-RegressivewitheXogenousinputneuralnetwork（NARX）;non-stationarytimeseries;windpower;totalpower摘要：探索构建对风电场总功率进行直接预测的高精度组合预测算法。考虑
