中央空調系統(tǒng)機組能效預測模型的研究.pdf

1、華中科技大學碩士學位論文中央空調系統(tǒng)機組能效預測模型的研究姓名：周靜娜申請學位級別：碩士專業(yè)：供熱、供燃氣、通風及空調工程指導教師：胡平放20110118II華中科技大學碩士學位論文AbstractAstheCOPoftheairconditioningsystemisanimptantindicationofsystemperfmancethispaperusesmodelingmethodstopredictCOPindertogi

2、veadvicesftheFaultdiagnosistoimprovethesystemefficiencyonthewholetosavetheenergy.IndertogettheoperationaldataofthechillerinabuildingduringtheairconditioningseasonthispaperusingtheenergyconsumptionanalysissoftwareEnergyPl

3、ustosimulatetheairconditionsystem.ThenusingtheBPneuralwktobuildtheCOPpredictionmodleoftheunitfinallyverifythereliabilityofthemodelingmethod.AstheaimofthispaperistopredicttheCOPoftheuinttheoutputofthemodelistheCOPatthet1m

4、emont.ByusingthemethodofregressionanalysisthefourparametersthatmostsignificantcrelationtotheCOParetheunitpowerthechilledwaterinletoutlettemperaturesthecoolingwateroutlettemperatureattmoment.Finallybydeterminingtheinputso

5、fthepredictionmodelaretheparametersmentionedabovebothattt1momentthebasicstructureofthemodelisfinished.Thispaperusesthreekindsofmethodstoimprovetheprecisionofthemodel.Firstistoincreasetheadaptivelearningratetoovercomethes

6、htcomingsoftheBPwk.Fexamplethesystemisunstableastrainingsometimetheresultsconvergetoalocalminimum.Secondistoadjustthefollowing3parametersthathavethebiggestinfluenceonthepredictionabilityofBPwkmodelthenumberofinputlayerno

7、deshiddennodesthemeansquareerr.Thirdistochangetheinputmothodsothatthemodelcanlearntheoperationruleofthewholeairconditioningoperationperiod.Byusingthemethodsmentionedabovethepredictionaccuracyisgreatlyimproved.Theabsolute

8、errsofexpectationswkoutputsaremostlyconcentratedin0.1to0.1intervalwhiletherelativeerrsaremostlyconcentratedin0.05to0.05interval.Toverifythereliabilityofthemodelmethodsmentionedabovethispaperusefollowingthreedifferentsyst