Abstract: Aiming at the no high accuracy of judging the wear state of diamond grinding wheel in engineering ceramics grinding, a fivelayer discrete wavelet decomposition of grinding acoustic emission (AE) signals was carried out based on the indepth study of the grinding mechanism of partially stabilized zirconia (PSZ) ceramics in the experiment of AE intelligent monitoring for precise grinding of PSZ with diamond grinding wheel. The results indicate that under the condition of severe wear of diamond grinding wheel, root mean squares and variances of the wavelet decomposition coefficients of the AE signals and the wavelet energy spectrum coefficients of the AE signals are increased at low frequency band than that of diamond grinding wheel slight wear. Using the 3 coefficients as the discriminative feature values of wear state of the diamond grinding wheel, classification accuracy of the diamond grinding wheel wear states in grinding PSZ is 100% by use of genetic algorithm support vector machine (GASVM), the classification accuracy by the GASVM is better than that by the BP neural network.
Key words: partly stabilized zirconia(PSZ) ; precision grinding；acoustic emission (AE); diamond wheel wear；wavelet analysis; genetic algorithm support vector machine(GASVM)