基于专家系统的鸡蛋品质分级技术研究-机械设计及理论专业论文.docx

StudyonClassifiedTechnologyforEggQualityBasedonExpertSystemAbstractEggappearanceparameterssuchasweightandsurfacecrackarethekey indexeswhichmeasuretheegg’Sexternalquality．弧ispapermakesuseoftheimage-processingtechnologytocollectparametersthroughanalyzingtheextemalappearancepictureoftheeggs，buildsupanexpertsystemtogradetheeggsbasedontheexternalappearancequality．田lemaincontentsareasfollows：Theexperimentdeviceisimprovedbasedontheoriginalexperiment．Throughadjustingtheparametersofthecameraandimageacquisitionsoftware，themoredearlyandlowerdistortionpicturesoftheeggsaregotten。Thecapturedeggimagesarepre-processedwiththeimage-processingtechnologytoextracttherelevantfeatureparameters，suchaseggshapeindex，weightandeggshellsurfacecrack．Onthebasisofthegroup’Sresearching,thetextcombinestheArtificialNeuralNetworktechnologyintoExpertSystembasedontheexternalappearancequalityaccordingtoeggsfeatureextractionandclassificationoftheindividualindicatorsalgorithmtheory．MakeuseofClanguagetocompiletheArtificialNeuralNetworkprocedure,thenmakeuseofsampledatatotrainnetwork，inordertobuilduptheknowledgedatabasethattheExpertsystemneeds．Finally,completethedesignoftheeggquali哆testingexpertsystem。BasedOilVisualC++6．0progrmnminglanguage,makeuseofprogrammingmethodwhichcatertoobjecttodesignaman-machineinterfaceofExpertSystemwhichbasedonNeuralNetwork。Thesystemiseasyfortheuserscarryouttheoperationofeggexternalqualityrating．Keywords：EggQuality,ArtificialNeuralNetwork，ExpertSystemMan—MachineInterfaceDirectedby：Prof．YUZhihongApplicantforMasterdegree：LIYang(MechanicalDesignandTheory)(CollegeofMachinery andElectricalEngineering,InnermongoliaAgricultmMUniversity,Hul氇ot010018，China)插图清单I．图1’前十位禽蛋生产大国2005年禽蛋产量对比。-。l2．图2机器视觉硬件系统示意图。53．图3图像采集软件界面64．图4不同光源下采集的鸡蛋图片75．图5专家系统结构图。86．图6神经元模型．1l7．图7S型函数128．图8前向网络．139．图9有反馈的前向网络．13lO．图lO层内有相互结合的前向网络1411．图1l相互结合型网络．1412．图12神经网络专家系统的基本结构图1913．图13通过知识．T程师获取知识2014．图14通过知识编辑器获取知识2015．图15多层前向BP网络．．2216．图16三层神经网络结构2517．图17神经网络训练误差2518．图18神经网络训练流程图．．2619．图19改动后的3层网络2720．图20新网络训练误差2