資源描述:
《組合預(yù)測模型構(gòu)建及其應(yīng)用》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、摘要論文題目:學(xué)科專業(yè):研究生:指導(dǎo)教師:組合預(yù)測模型的構(gòu)建及其應(yīng)用應(yīng)用數(shù)學(xué)劉素兵王秋萍副教授要簽名:塑!壹幺簽名:蘭盤壟隨著灰色系統(tǒng)理論、BP神經(jīng)網(wǎng)絡(luò)等新的理論技術(shù)應(yīng)用于預(yù)測領(lǐng)域,預(yù)測技術(shù)得到了很大的發(fā)展。由于灰色預(yù)測模型對一般的預(yù)測模型具有很強(qiáng)的融合力和滲透力,將灰色模型與其它模型結(jié)合進(jìn)行分析和預(yù)測,可以提高預(yù)測精度。因此,本文對灰色模型與其它模型的結(jié)合進(jìn)行了分析和研究,建立了組合預(yù)測模型,并對中國能源消費(fèi)量進(jìn)行了預(yù)測。本文的主要研究內(nèi)容和成果如下:1.提出了基于灰色預(yù)測GM(I,1)模型、三角模型和時(shí)間序列分析ARMA模型的組合預(yù)測摸型TGMA(I,1)。該模型以灰色預(yù)測GM(I,1)模
2、型擬合數(shù)據(jù)序列的趨勢項(xiàng),以三角模型和ARMA模型捕獲系統(tǒng)的殘差序列。2.構(gòu)建了灰色神經(jīng)網(wǎng)絡(luò)組合預(yù)測模型。使用BP神經(jīng)網(wǎng)絡(luò)對三種改進(jìn)的灰色預(yù)測模型(灰色優(yōu)化預(yù)測模型GOM(I,1)、無偏灰色預(yù)測模型、改進(jìn)的新陳代謝預(yù)測模型)迸行組厶口o3.對灰色神經(jīng)網(wǎng)絡(luò)組合預(yù)測模型做進(jìn)一步的改進(jìn)。采用灰色關(guān)聯(lián)分析法找出對中國能源消費(fèi)量有較大影響的幾個(gè)因子,以它們的時(shí)間序列數(shù)據(jù)作為BP神經(jīng)網(wǎng)絡(luò)的輸入,這樣全面綜合地考慮到能源消費(fèi)系統(tǒng)的影響因素,從而提高了預(yù)測精度。關(guān)鍵詞:組合模型;灰色預(yù)測:灰色關(guān)聯(lián)分析;ARMA模型;神經(jīng)網(wǎng)絡(luò)模型hbstractTitle:THEESTABLISHMENTANDAPPLICAIT
3、lONOFCOMBINAITIONFORECASTINGMODELMajor=AppliedmathematicsName:SubingLIUSupervisor=Associateprof.QiupingWangAbstractSignatureWiththeapplicationofthegreysystemtheoryandtheBPneuralnetworkinthefieldofforecasting,theforecastingtechnologyhasobtainedthegreatdevelopment.Thegreyforecastmodelhastheverystrongf
4、usionstrengthandpenetrabilitytothegeneralmodel,SOitCanbecombinedwithothermodelstoimprovetheforecastingprecision.Therefore,analysisandresearchofthegreymodelwithothermodel’Scombinationarecarried,thecombinationforecastingmodelsareestablishedandtheenergyconsumptionofChinaisforecasted.Theprimaryresearchc
5、ontentandresultsareobtainedasfollowes:1.BasedonthegreyforecastmodelGM(1,1),thetrianglemodelandtheARMAmodel,thecombinationforecastmodelTGMA(1,1)hasbeenproposed.GM(1,1)isusedtofitthetendencyitemofthedataseries,meanwhile,thetrianglemodelandARMA(p,q)modelcapturesystem'sresidualsequence.2.Thegrey-neuraln
6、etworkcombinationmodelhasbeenbuilt.Threeimprovedgreymodels(thegreyoptimizationmodelGOM(1,1),theagonicgreyforecastmodelandtheimprovedmetabolismforecastmodel)arecombinedthroughthetheoryoftheneuralnetworkmodel.3.Thefurtherimprovementofthegrey-nerualnetworkcombinationmodelhasbeenmade.Thegreyrelationanal
7、ysismethodisadoptedtofindthemaininfluencingfactorsoftheenergyconsumptionandtimeseriesdataofthefactorsareusedasaninputofBPneuralnetwork.So,theenergyconsumptionsystem’Sinfluencingfactorsalealsoconsidere