轴承故障, 谐波小波变换, 奇异值分解, 信息熵, 实时诊断

A new method for real-time diagnosis of bearing faults was proposed based on harmonic wavelet, SVD and entropy. The vibration signals were transformed by using harmonic wavelet to get harmonic wavelet coefficients. The wavelet coefficients were used to construct harmonic wavelet time-frequency distribution matrix, scale for rows and time for columns. The singular values were obtained through time-frequency matrix SVD. The information entropy was computed in the partition criteria of singular values to diagnose bearing faults. A concrete model based on harmonic wavelet singular entropy was presented. Through many inner- outer ring fault experiments, the results show that the method is effective to analyze bearing faults, has high degree in real-time, which achieve real-time diagnosis in less 68kHz sampling frequency, has good applicability.