鹈鹕算法优化VMD参数与RF的滚动轴承故障诊断.pdfVIP

机械工程师

MECHANICALENGINEER

鹈鹕算法优化VMD参数与RF的滚动轴承故障诊断

纪佳呈

（沈阳化工大学机械与动力工程学院，沈阳110142）

摘要：为更好地提取滚动轴承故障特征以及提高故障检测的准确性，建立一种以鹈鹕算法优化变分模态分解（VMD）参数

联合随机森林算法的故障诊断模型。首先为解决VMD算法中人为经验选取核心参数可靠性低这一问题，引入鹈鹕优化算法

（POA），将最小包络熵作为适应度函数，自动搜索理想参数；其次，提取轴承各状态的多种信号特征，构建特征向量，获得重要性排序；

最后将其放入随机森林（RF）分类器中，实现滚动轴承故障识别。结果表明，该方法在信号的自适应分解、有效故障信息的提取和

故障分类方面具有较高的可行性，故障识别率为99.524%。

关键词：变分模态分解（VMD）；鹈鹕优化算法（POA）；随机森林（RF）；轴承故障诊断

中图分类号:TH133.3文献标志码:粤文章编号：员园园圆原圆猿猿猿（圆园24）07原园059原园5

RollingBearingFaultDiagnosisBasedonPelicanOptimizationAlgorithmVMDParametersandRF

JIJiacheng

(SchoolofMechanicalandPowerEngineering,ShenyangUniversityofChemicalTechnology,Shenyang110142,China)

Abstract:AfaultdiagnosisapproachbasedonthePelicanoptimizationalgorithmandthevariablemodedecomposition

(VMD)parameterscombinedwiththerandomforestalgorithmisproposedinordertobetterdistillthefaultfeaturesofrolling

bearingsandincreasetheexactitudeoffaultdetection.First,thepelicanoptimizationalgorithm(POA)ispresentedtoaddressthe

issuethatitischallengingtoestimatethenumberofmodalcomponentsandpenaltyfactorsinVMD,andtheminimum

envelopeentropyisutilizedastheobjectfunctiontoautomaticallysearchfortheidealparameters;Second,severalbearing

statesignalfeaturesareextracted,featurevectorsarebuilt,andimportancerankingsaredetermined;Lastly,itisfedintoa

randomforest(RF)classifiertoenabletheidentificationofrollingbearingfaults.Thefindingsdemonstratethatthismethodis

highlyfeasibleinsignaladaptivedecomposition,successfulinextractingfaultinformationandfaultclassification,andthatthe

fau