資源描述：
《基于遺傳神經(jīng)網(wǎng)絡(luò)的汽車故障率預(yù)測》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、基于遺傳神經(jīng)網(wǎng)絡(luò)的汽車故障率預(yù)測上海交通大學(xué)碩士學(xué)位論文基于遺傳神經(jīng)網(wǎng)絡(luò)的汽車故障率預(yù)測姓名：楊婷中請學(xué)位級別：碩士專業(yè)：控制理論與控制工程指導(dǎo)教師：楊根科20080101上海交通大學(xué)碩士學(xué)位論文i基丁遺傳神經(jīng)網(wǎng)絡(luò)的汽車故障率預(yù)測摘要汽車在使用過程屮總會有故障出現(xiàn)，對于汽車生產(chǎn)商來說，了解到汽車在一定行駛條件下的故障率，可以減少相應(yīng)的時間成本和庫存成本。有鑒于此，本文首先通過對汽車整體性能的分析，找出可能引起整體性能指標降低而導(dǎo)致汽車故障的因素：汽車不同零件的易損度不同，零件本身質(zhì)量差異，汽車維修頻率，汽車消耗品種類，汽車行駛環(huán)境，駕駛因素，行駛距離等，開拓了汽車故障模型建
2、模的新途徑;并對每個不同的故障因素影響導(dǎo)致汽車故障的權(quán)重進行了分析評估，得出汽車故障率的控制模型。其次，本文采用了GA-BP神經(jīng)網(wǎng)絡(luò)處理問題。在描述了傳統(tǒng)BP網(wǎng)絡(luò)的基本模型的基礎(chǔ)上，介紹了用于BP網(wǎng)絡(luò)中的常見的幾種改進算法：附加動量和學(xué)習(xí)率自適應(yīng)調(diào)整的改進BP算法BPX、Levenberg-Marquardt優(yōu)化方法LM>Bayes規(guī)范化BP算法。并利用遺傳算法對神經(jīng)網(wǎng)絡(luò)的權(quán)值和閾值進行優(yōu)化，在MATLAB上分別對這幾種算法對模型進行了訓(xùn)練，利用神經(jīng)網(wǎng)絡(luò)的泛化能力，得到我們所需要的預(yù)測數(shù)據(jù)。關(guān)鍵字：BP神經(jīng)網(wǎng)絡(luò)、遺傳算法、優(yōu)化、汽車故障率、預(yù)測1上海交通大學(xué)碩士學(xué)位論文iF
3、ORECASTTHECARFAILUREBASEDONBPNEURALNETWORKABSTRACTTngeneral,theautomobilehasthebreakdownappearanceintheuseprocess.Regardingtheautomobileproducer,itwouldreducethecorrespondingtimecostandinventorycostafterknowingtheautomobilecondition.failurerateundercertaintravelAstothis,firstly,throughthea
4、utomobileoverallperformanceanalysis,thearticlefindsthepossiblecausestoreducetheautomobilebreakdown,suchastheautomobileeachinsiallmentbuckle,thecomponentsqualitycliffcrcnccs,theautomobileservicefrequency,theautomobilefueloil,theautomobiletravelenvironment,thedrivingtechnologyandthedrivingme
5、thod,thetravelcourseandsoonandetc.Allofthesehavedevelopedtheautomobilebreakdownmodelinthenewway,andanalysistheweighofthedifferentbreakdownfactorsinfluenee,thentheautomobilefailureratecontrolmodelhasbeengot.Next,thisarticlehasusedtheGA-BPneuralnetworktosolvethequestion.Onthefoundationofthet
6、raditionalBPnetworkbasicmodel,itintroducestheseveralcommonkindsofimprovementalgorithmintheBPnetwork:Additionalmomcntumandlearningrateauto-adaptcd■1上海交通大學(xué)碩士學(xué)位論文iiadjustmentimprovementBPalgorithmBPX、Levenberg-MarquardtoptimizedalgorithmLM,BayesstanclarclizcclBPalgorithm,andcarriesontheoptimi
7、zationontheneuralnetworkweightandbiasusingthegeneticalgorithm.IthasseparatelycarriedonthetrainingmodelbyMATLABonthebasisofthesealgorithms,andforecastthedatawhichweneedthroughregressionabilityusingtheneuralnetwork.KEYWORDS:Back-PropagationNeuralNetwork,GeneticA
