Tool in the process of grinding tend to inspire a large number of voice signal, the signal contains abundant information tool, by means of cutting tool grinding process monitoring voice signal processing and analysis, to understand the tool grinding process, improve the stability of the preparation of existing tools.
Because of the sound signal is vulnerable to the external environment disturbance, you first need to single experiment was carried out for the external environment factors and this helps to understand the environmental factors, also is helpful for de-noising in the tool sharpening signal analysis and frequency separation. Sampling background noise reciprocating swing back and forth, grinding wheel, grinding wheel rotation noise and cooling fluid noise, the acquisition of sound signal analysis of the power spectral density, can be more clearly characteristic frequency of various environmental factors, among them, grinding wheel rotation noise is the largest.
On this basis, carried out on a high speed steel cutting tools in good condition, grinding, gathering the voice signal in the process of grinding. Then do artificial defects on the cutting tool, grinding again, also collecting sound signal, and the contrast study on the two sound signal. Power spectral density, it is found that grinding wheel rotation noise impact on the acquisition of tool grinding sound signal is very large. Comprehensive consideration, choose Sym6 wavelet base, to 5 layer signal wavelet packet decomposition, choose the appropriate node to separate the noise, and then the reconstruction, and to reconstruct the signal power spectral density analysis. This method can be submerged in the environmental noise tool grinding sound signal extracted, through comparative study, found that characterization tool grinding characteristics of frequency distribution in high frequencies, while the intact and damaged in the spectrum of tools of the frequency spectrum of wave shape and the same proportion to each other, but different, and the damaged knives on 7.2 ~ 10 kHz frequency spectrum also appeared a lot of active peaks. Will get the original signal acquisition and signal by wavelet packet processing as Wave audio format file, comparing with the real sound signal in the process of cutting tool grinding, confirmed the basic conform to the real voice heard while sharpening, basic suitable signal reconstruction method.
HSS tool sharpening, relative to the environmental factors in the process of collecting the tool sharpening a weak voice, directly on the data processing, the effect is not ideal, using wavelet/wavelet packet decomposition of signal, signal node selection contains useful information, and then the refactoring, analysis processing, can well solve the problem of weak signal extraction, the processing method was applied to signal monitoring tool sharpening process is feasible.