基于VMD-CWT和SwinTransformer的滚动轴承故障诊断方法

打开文本图片集

中图分类号：TH133.33 文献标志码：A 文章编号：1671-5276（2025）06-0018-06

Abstract：Inresponsetothenoiseinterferenceandinaccuratefaultfeatureextractioninrolingbearing faultsignals，an inteligentfaultdiagnosismethodforrolingbearingscombiningvariationalmodedecomposition（VMD），continuouswavelet transform（CWT），andthe Swin Transformernetwork isproposed.Thesignalisdenoisedusing variational modedecomposition， andthe econstructed signal isconverted into a time-frequency map using CWT.The two-dimensional feature maps obtained fromthetime-frequencymapsareusedasinputs totraintheSwinTransformermodel，achieving intellgentfaultdiagnosisof roling bearings.TheexperimentalresultsindicatethatthemethodcombiningVMD-CWTwith Swin Transformer network has better fault diagnosis accuracy，with the accuracy of the test set in the measured data up to 99.79%

Keywords：rolingbearing；variationalmodedecomposition；continuous wavelettransform；Swin Transformer；faultdiagnosis

0 引言

随着工业设备的广泛应用，滚动轴承在各种旋转机械设备中起着不可或缺的作用，准确识别轴承故障类型并及时采取有效维修措施至关重要。（剩余8237字）