Research on the Signal De-noising Method of Acoustic Emission in Fused Silica Grinding

Abstract

The ultra-precision grinding process of brittle and hard fused silica is very complex. In order to monitor the grinding process accurately, it's necessary to de-noise the acoustic emission (AE) signals generated in this process and extract useful parameters which can characterize the cutting procedures of abrasive grain. Firstly, according to the characteristics of AE signal when single diamond grain scratching, the AE signal with white Gaussian noise during grinding process was simulated, whose SNR was below -2dB. Then the simulated AE signal was de-noised by wavelet threshold de-noising method, empirical mode decomposition (EMD) threshold de-noising method and EMD-Wavelet threshold de-noising method. Taking the signal to residual noise ratio (SRNR) and the mean square error (RMSE) as the evaluation parameters, the optimal way was EMD-Wavelet threshold de-noising method. The SRNR increased to 9dB, and the RMSE reduced to 0.017. At the end, the AE signal acquired from fused silica grinding process was de-noised by the optimal method, and the cutting process of the abrasive particles can be observed accurately. Taking the number and energy of impulse oscillation per unit time as key parameters, the accurate monitoring of the grinding process of fused silica material was realized.