基于全矢-AR模型的旋转机械故障趋势预测方法分析

《基于全矢-AR模型的旋转机械故障趋势预测方法分析》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、AbstractForfaulttrendpredictionoflarge-scalerotatingmachinery,thetraditionalARpredictionmodelisbasedonsingle-sourceinformation.Infact,duetothewhirlingmotioncharacteristicsoftherotor,thestructureofthefrequencyspectrumatthesamecrosssectionisdifferentinthedifferentdirections.Asaresult,accordingtofa

2、ulttrendpredictiontosingle-sourceinformation,theresultisdifferent.Toensurethereliabilityanduniquenessofthepredictingresult,fullvectorARpredictionmodel,combiningwithfullvectorspectrumtechnologyandtheARpredictionmodel,isconstructed.BecauseARmodelandfullvectorARmodelhavethecharacteristicsofhighpred

3、ictionaccuracyinshorttermandlowpredictionaccuracyinlongterm,theKalmanfilteringisintroducedintoARmodelandfullvectorARmodeltoimprovethepredictionaccuracy.Finally,theireffectivenessisverifiedbyexamples.Themainresearchworkisasfollows:(1)ResearchonARModelbasedonfullVectorSpectrumanditsapplicationinth

4、efaulttrendpredictionoftherotorsystemoflarge-scalerotatingmachinery.Thespecifictheoreticalcalculationformulaanditscorrespondingfaulttrendpredictionflowchartaregiven.TheexperimentalresultsshowfullVectorARModelcanpredictfaulttendencyintherotorsystemoflarge-scalerotatingmachineryeffectivelyandensur

5、ethepredictionresult'sreliabilityeffectively.(2)ResearchonARModelbasedontheKalmanFilteranditsapplicationinthefaulttrendpredictionoftherotorsystemoflarge-scalerotatingmachinery.TheAR-Kalmanprediction(ARKF)method,combiningwiththeKalmanFilterandtheARmodel,isproposed.Thespecifictheoreticalcalculatio

6、nformulaanditscorrespondingfaulttrendpredictionflowchartaregiven.Theexperimentalresultsshowtheproposedmethodcanpredictfaulttendencyoflarge-scalerotatingmachineryeffectivelyandimprovethelinearpredictionaccuracyoftraditionalARmodelobviously.(1)ResearchonfullvectorARModelbasedontheKalmanFilterandit

7、sapplicationinthefaulttrendpredictionoftherotorsystemoflarge-scalerotatingmachinery.ThefullvectorARKalmanforecast(FARKF)ModelandthefullvectorARKalmancorrection(FARKC)Model,combiningwithfullvectorspectrumandARmodelandtheKalma