Rolling bearing fault diagnosis method based on mean singular value screening

For the problem that the vibration signal of aeroengine main bearing is affected by strong background ambient noise and aiming at achieving accurate diagnosis of rolling bearing faults, a rolling bearing fault diagnosis method based on mean singular value screening is proposed. Firstly, the vibration signal is decomposed into eight signal components by WPT (Wavelet packet transform), then arrange the components into a matrix with eight row and process it through SVD (singular value decomposition). The noise filtering of vibration signal is realized by the method of threshold filtering so that fault features hidden in strong background noise can be accurately extracted through envelope demodulation at last. According to simulated signal experiment, the signal-to-noise ratio of the denoised signal is increased by 7.55 dB which verifies the noise processing effect of the method, and after that, the theoretical feasibility and practical engineering application value are verified by carrying out experiments on rolling bearings under the condition of complex transmission paths and real aeroengine.