基于全局优化预测的温室智能控制模型-程曼

11211．0710012．100083、。、、、。。S625．5TP181A1003－188X201310－0026－040、、、、1。、、。2。。3。、、[e-mail]。。。1。、。。4。1．1、、、、、、。5。1．2AIArtificialIntelligence。1。·62·201310106。。。2。。、、。296m2161。5712345。1CO2。1Fig．1Thethermalzonestructureofexperimentgreenhouse1Table1Datacollectioncategoryandquantityofzones11002161603201410051013、。3。2。2Fig．2Intelligentcontrolstructureofgreenhouse3．13．1．1。。3．1．2。1。。。TＲ=α0+α1TL+α2ＲL+α3Ｒ2L1TＲ—TLＲL—αi—αi。3．1．34621、16·72·20131010、2、6、1。。41234、、。3．2。3。3232。3．33．3．1BP7、、、。BPBP。BP、3。1。。8。3。3BPFig．3TopologyofBPneuralnetwork3ijkWijTjk。XiYkOj=f∑mi=1Xi×W()ij2Yk=f∑nj=1Oj×T()jk301Sigmoidfx=11+e－x4E=12∑ti－Oi25ti—iOi—i。ΔWijn+1=h×Φi×Oj+ΔWijn6h—Φi—iOj—j。3．3．2000－2400124。4201135。。4Fig．4Temperatureandhumiditysimulation·82·2013101041。2、、、。1．PLCJ．2010383419827－19828．2ZazuetaFSBucklinＲJonesPHetal．BasicconceptsinenvironmentalcomputercontrolofagriculturalsystemsC//AgriculturalandBiologicalEngineeringDeptInstituteofFoodandAgriculturalSciences．FloridaUniversityofFlor-ida2008．3BlascoXMartinezaMHerreroaJMetal．Model－basedpredictivecontrolofgreenhouseclimateforreducingenergyandwaterconsumptionC//ComputersandElectronicsinAgriculture．AmsterdamElsevierSciencePublishersB．V2007．4．PIDJ．2011272307－311．5．J．2004201246－249．6．D．2001．7PEredicsTPDobrowiecki．HybridknowledgemodelingforanintelligentgreenhouseC//2010IEEE8thInternationalSymposiumonIntelligentSystemsandInformatics．Serbia459－463．8．M．2006．TheModelinIntelligentControlofGreenhouseBasedonGlobalOptimizationPredictionChengMan1YuanHongbo12CaiZhenjiang11．CollegeofMechanicalandElectricalEngineeringAgriculturalUniversityofHebeiBaoding071001China2．CollegeofInformationandElectricalEngineeringChinaAgriculturalUniversityBeijing100083ChinaAbstractAbstractThepaperdealswiththeproblemofmodelingandcontrolofgreenhousesinsideclimatebasedontheglobalvariableoptimizationmethodforprediction．Amathematicalmodelofgreenhouseclimatewasestablished．Con-frontedwithproblemofgreenhouseclimatecontrolexistedinconventionalcontrollersuchascontrolsystemisreactivetheadjustmentoftheactuatorsisnotsynchronizedcontrolschemeisnotoptimal．Inthemethodinsidethegreenhousetemperaturehumidityradiationvaluescropgrowthstatuscurrentstateofactuatorsexternalenvironmentandthelocalweatherconditionsbydatafusionasthe－globalvariables．Thenthegreenhouseofthefuturestateoftheenvironmentshort－termpredictivevalueareobtainedbymathematicalmodelofneuralnetworkcontrol．Thesimulationresultstestifythevalidityandreasonabilityoftheglobaloptimizationpredictioncontrolstrategyfortheclimatecontrolinthegreen-houseandtheachievementhascertainreferencevalueforthedevelopmentinintelligentcontrolofthegreenhousemicro-climate．Keywordsgreenhouseintelligentcontrolglobalvariableoptimizationpredictionneuralnetwork·92·20131010