For the nonstationary and nonlinear signals generated by the flexible thinwalled bearings in the event of failure,the Hilbert-Huang transform (HHT) method is used to analyze them.For the problem that the noise signal will pollute the entire Hilbert spectrum,a PCA noise reduction algorithm based on SVR spectrumis proposed.Through simulation analysis and flexible thinwall bearing failure experiments,from the Hilbert spectrum of the bearing inner and outer ring fault simulation signals and actual signals,it can be concluded that the change rate of the outer ring fault signal frequency is twice that of the bearing frequency,the change rate of the inner ring fault signal frequency is consistent with the bearing frequency.This conclusion can provide theoretical guidance for the practical application of the project.