Bearing fault diagnosis based on symplectic geometry extraction transformation

Abstract:  Aiming at the problem that the vibration signal of rotating machinery was noisy and the fault features were difficult to extract, a time frequency analysis (TFA) method based on symplectic geometry extracting transform (SGET) was proposed and applied for fault diagnosis of rolling bearing. First, the symplectic geometric similarity transformation was used to solve the eigenvalues of the Hamiltonian matrix, and then it was used to reconstruct multiple single-component signals, the effective component was selected and used to reconstruct the original signal. The synchro extraction operator (SEO) was introduced into the reconstructed signal to eliminate the divergent energy in the time-frequency signal,and obtain clearer time-frequency information. Finally, the effectiveness and applicability of this method are verified by simulation and experiment, and compared with other classical methods. Simulation and experimental results show that the Renyi entropy values of the inner and outer rings are respectively 13.862 and 15.068, which are lower than the other three methods, can effectively diagnose the bearing fault frequency and its one, two and three times of frequency, get clearer while effectively reducing noise concentrated time-frequency energy representation, and enhanced robustness under strong noise background. The research results show that the method can effectively denoise the bearing fault vibration signal and extract the fault features for analysis.

Key words:  rotating machinery; rolling bearings; fault diagnosis; symplectic geometry extracting transform (SGET); time-frequency analysis(TFA);feature extraction