Aiming at the shortcomings of traditional mechanical fault detection methods based on Shannon-Nyquist theory, such as high sampling rate and low fault detection rate, a mechanical fault signal detection method based on compressed sensing was proposed. The random Gaussian measurement matrix was constructed based on the basic theory of compressed sensing, and the dimension of the original fault signal was reduced by compression after noise reduction. The sparse solution of the compressed sensing matrix was obtained by the norm sparse approximation method. The time domain and energy features of the original compressed fault signal were extracted to reconstruct the original fault signal. The simulation results show that the signal reconstruction effect of the proposed fault detection method is better, and the detection rate of the fault signal can reach 96.16%.
LU Hongrong. Research on Mechanical Fault Signal Detection Method Based on Compressive Sensing[J]. Machine Tool & Hydraulics,2021,49(4):183-188