基于davinci平台的除尘风机自动控制系统

《基于davinci平台的除尘风机自动控制系统》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、浙江大学硕士学位论文统对烟气图像特征提取和烟气浓度分级模型分类算法进行了实现和优化；并对控制系统效果进行了测试。关键词：节能，自动控制，DaVinci，特征提取，多分类器模型IVAbstractAlotofabandonedsmokeisproducedduringthescrapcoppersmeltingprocess．Inordertoreducetheairpollution，itisnecessarytousehighpowerdustfantooptimizetheoperation，whichcandrivesmok

2、etopurpose-builtbaginrecycle．Tosolvetheproblemthatavailabletechnologyisdifficulttodetectsmokeconcentration，anautomaticcontrolsystemofdustfanisdevelopedbaseonimageanalysisinthispaper，whosecoreisTMS320DM6446(DaVincifamily,ARMandDSPdouble‘coreprocessor)．Imagedfeaturesrel

3、atedtosmokeconcentrationpropertyareextractedfirstlyfromthesmokeimage，whichiscapturedbytheindustrialcameraofthesystem．Then，gradesmodelofsmeltingprocessandsmokeconcentrationlevelbasedonPSFCMandFCMisdeveloped，eachofwhichmapstooptimalspeedofdustfan．Finally,therotationalsp

4、eedisadjustedinrealtimebythedistributionofdataofgraphicfeature．Thecontrolsystemisappliedinasmeltingplantandtheresultshowsitsvalidaty，stabilityandeffectiveness．Themaincontributionsandinnovationsofthethesisinbriefareasfollowing：(1)Designschemeofthecontrolsystem，anddevel

5、opmentofDaVincidual‘corehardwareplatform．Accordingtothecharacteristicsofsmoke，feedbackcontrolschemewithfeedforwardisdesigned．Thenthecircuitsarecompletedbasedonmodulardesignconcept．Withahighdegreeof。flexibilityandusability,platformconsistsofanimagacquisitionunit，animge

6、outputunit，thepowerssupplycircuit，genericpedpheraicircuitandeta1．(2)Smokeimagecharacteristicvaluesextraction．Throughthecraftresearchofsmeltingprocess，amethodwhichcombinesgrayfeature,texturefeatureandmovementfeatureofsmokeimageisforwardtomeasureandestimatetheconcentrat

7、ionofsmoke，andthe：feasibilityandalgorithmisanalysedlater．(3)Researchandreafizationofsmokeconcentrationgradesmodel．Assometypicalfeaturessampleofsmokeimagescanbelabeled，smokeconcentrationgradesV浙江大学硕士学位论文modelissetupbySVMmulti—classifier,whichprovesitsfeasibilitytobuild

8、modelbasedonimgefeatures．Duetocomplexalgorithm，incompletemodel，andthedifficultlabeling，allinnovativeviewpointbasedonPSFCMand