Synchroextracting transform (SET) can obtain the ideal time-frequency spectrum by extracting the time-frequency coefficient of the short-time Fourier transform (STFT) at the immediate frequency position. On the one hand, the method improves the time-frequency resolution, on the other hand, it reduces the influence of the cross terms, and suppresses the interference of the noise on the spectrum of the STFT to some extent. However, the reconstruction of signal components and the application of fault diagnosis based on the SET spectrum are still to be further studied.This paper proposes a SET signal component extraction and reconstruction method based on order statistics filter (OSF).Firstly, the marginal spectrum is used to characterize the amplitude of the signal in the spectrum when the SET changes with frequency in the whole frequency band, and then the margin spectrum is divided by the order statistics filter. Each of the signal components is reconstructed by the SET inverse transform after mapping the segmentation obtained boundary to the SET time spectrum. The shock-sensitive kurtosis index is used to screen the components of each component that may contain faults, and the effectiveness of the proposed method is demonstrated by the simulated signal and the inner ring fault signal of the rolling bearing.