大機組振動狀態(tài)預(yù)報技術(shù)的研究開發(fā)與應(yīng)用.pdf

1、作指系專研大機組振動狀態(tài)預(yù)報技術(shù)的研究開發(fā)與應(yīng)用者：塹壹壘導(dǎo)老師：望燕堡塾蕉別：垡蘭堂墮業(yè)：垡鱟墨堡究方向：垡王遮查些型魚壘斷論文評閱人：答辯委員會主席：申請學(xué)位級別：三壁亟學(xué)位授予單位：塑鋈盤堂論文提交日期：二00四年九月AbstractThisthesisfocusesOilconditionforecastingtechniqueoflargemachine—sets，whichmanyenterprisesiscryingfor

2、AnoptimizedforecastingmethodbasedOilartificialneuralnetwork(ANN)isdeveloped，andthenitisputintopracticeTheresultshowsthatthismethodiSsuitableforconditionforecastingofmachineryFirst，varioustraditionalforecastingmethodsarei

3、ntroducedinbriefButthesemethodsareunfitted船non—linearobjectsANNhasmanyadvantagesoverthesetraditionalmethodsItispointedoutthatmachine—setsconditioncanbepredictedbysuchnon—linearmethodsasANNTIlen，thisthesisdiscusseshowtoes

4、tablishtheANN’Sforecastingmodel。Somekeyquestionsareanalyzed。suchaSdeterminingANN’Stopologicalstructure，selectingANN’Strainingaritlunetic，andsoon、AfterestablishingtheforecastingmodelANNisintroducedtoforecastmachineryopera

5、tionconditionFirstly，theconceptofpredictivestabilityispresentedThemainfactorsaffectingANN’Spt。edictiveperformancearediscussedTheresultdemonstratesthattheANN’Spredictiveaccuracyisdependedontheamptitudeofnoiseintrainedsamp

6、lesSecondlybasedontheconceptofffactaldimension，anindexispresentedtoevaluatewhetheratrainedsamplesisdivinableThirdlybased0ntheconclusionabox’e，anewoptimizingapproachisputforward，Experimentsandapplicationsprovethatthismeth