Signal de-noising in gear pitting fault identification by an improved singular value decomposition method

Abstract

AbstractIn this research a new method of improved singular value decomposition (ISVD) is proposed for the vibration signal de-noising of gear pitting fault identification. In this method, the delay time τ and embedding dimension m of the Hankel matrix for SVD are optimized by autocorrelation function and Cao’s algorithm respectively. Simulation and experiments are conducted to demonstrate the method. In the simulation, the ISVD method is employed to de-noise the artificial vibration signal in a mathematical model of gear pitting fault, the result demonstrates the signal-noise ratio (SNR) value is SNR = 31.3 dB, and the root-mean-square error (RMSE) value is RMSE = 0.34. In the experiment, the ISVD method is adopted to de-noising the vibration signal of gear pitting fault identification, the results demonstrate SNR is SNR >45 dB, and the RMSE value is RMSE <0.4 of the fault characteristic signals at each measuring point position. The results of simulation and experiment show, the ISVD method is efficient to de-noise the vibration signal of gear pitting fault.