資源描述:
《vd爐鋼水溫度預(yù)報模型開發(fā)》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、東北大學(xué)碩士學(xué)位論文VD爐鋼水溫度預(yù)報模型開發(fā)姓名:周勇申請學(xué)位級別:碩士專業(yè):計算機應(yīng)用技術(shù)指導(dǎo)教師:呂振遼20040101查!!壟堂壁主堂堡壘查一一—立i:塾VD爐鋼水溫度預(yù)報模型開發(fā)摘要建立一個VD爐鋼水溫度的預(yù)報模型,在了解和掌握VD處理過程中鋼水溫度變化規(guī)律的基礎(chǔ)上,制定合理的溫度制度,實現(xiàn)對VD終點溫度的精確控制,對生產(chǎn)的順行有著重要的意義。本文通過對現(xiàn)場的系統(tǒng)結(jié)構(gòu)、網(wǎng)絡(luò)的分析,設(shè)計了VD溫度預(yù)報系統(tǒng)。通過遠(yuǎn)程訪問現(xiàn)場的ORACLE數(shù)據(jù)庫,提取基礎(chǔ)數(shù)據(jù):通過對工藝的分析,提取神經(jīng)網(wǎng)絡(luò)程序處理的數(shù)據(jù);設(shè)計了本地
2、數(shù)據(jù)庫,實現(xiàn)對數(shù)據(jù)的合理存儲。VD溫度預(yù)報系統(tǒng)程序由三個應(yīng)用程序組成:1.接口程序;2.主程序;3.神經(jīng)網(wǎng)絡(luò)程序。接口程序為系統(tǒng)提供現(xiàn)場數(shù)據(jù),保存現(xiàn)場數(shù)據(jù)。主程序?qū)崿F(xiàn)對數(shù)據(jù)的整理、篩選,為神經(jīng)網(wǎng)絡(luò)程序提供數(shù)據(jù)文件,并將預(yù)測的結(jié)果以圖形的方式顯示,保存訓(xùn)練的結(jié)果。神經(jīng)網(wǎng)絡(luò)程序?qū)崿F(xiàn)對VD過程溫度的預(yù)測,并且能夠?qū)ψ陨淼纳窠?jīng)網(wǎng)絡(luò)進(jìn)行訓(xùn)練,提高預(yù)測精度。系統(tǒng)分成了三個部分,使現(xiàn)場數(shù)據(jù)的提取過程與數(shù)據(jù)的處理過程分離,增強了系統(tǒng)的靈活性;獨立的神經(jīng)網(wǎng)絡(luò)程序使得神經(jīng)網(wǎng)絡(luò)的結(jié)構(gòu)可以不斷的改進(jìn)。系統(tǒng)不僅實現(xiàn)了對VD終點溫度的預(yù)報,而且實現(xiàn)
3、對VD高真空過程中溫度的預(yù)報,對精確控制VD時間有著重要的意義。關(guān)鍵詞VD(VacuumDegas)溫度預(yù)報神經(jīng)網(wǎng)絡(luò)數(shù)據(jù)庫數(shù)據(jù)分析一Ⅱ一查!!查蘭堡主堂堡壘查壘里!!墾竺!PredictorSystemofMoltenSteelTemperatureinVDStoveAbstractAfterunderstandingandmasteringchangingtemperatureruleofmoltensteeldealtwithinVDstove,createapredictortemperaturemodelofmo
4、kensteelinVDstoveandmadeareasonedtemperaturesystemandcontrollasttemperatureofmoltensteelinVDstoveaccurately,thosearesignificativetoproducesteel.DesignPredictorSystemofmoltensteeltemperatureinVDstove,afteranalyzingsystemandNetworkstructureofequipment.GetBasicDatao
5、fproducefromOracleDataBasebylong-distanceNetworkandNeuralNetworkProgramDatabyanalyzingtechniques.InordertostoreallData,setuplocalDataBase.PredictorSystemofmoltensteeltemperatureinVDstoveismadeupofthreeappliedProcedure,namely:AccessProcedureandMainProcedureandNeur
6、alNetworkProgram.AccessProceduregetsBasicDataofproduceandstoreDateatsametimeMainProcedureOfferpredictorDatafilesandtrainfiles,aRerBasicDataofproducearesealedandfilteredbyMainProcedure.PredictorresultsareshowedwithgraphicsmodeandtrainresultsaresavedbyMainProcedure
7、.NeuralNetworkProgramdopeoutmoltensteeltemperatureinVDstove,NeuralNetworkProgramtrainthemselvesPrecisionsofpredictorresultsareimproved.AccessProcedureandMainProcedureandNeuralNetworkProgramconstitutePredictorSystemofmoltensteeltemperatureinVDstove,itisseparatedth
8、atPredictorSystemofmoltensteeltemperatureinVDstovegetsandhandlesDataofproduce.So,itisagilethatNeuralNetworkProgramstructurecallbeimprovedbydependentNeuralNetwo