<![CDATA[Monitoring of metal cutting tool wear states is a key technology for automatic, unmanned and adaptive machining. As tool wear increases, the vibration signals of cutting tool become more and more irregular in the turning processes. The degree of tool wear can be indirectly monitored according to these changes of vibration signals. In order to quantitatively describe these changes, fractal theory and fuzzy clustering method were introduced into the cutting tool wear monitoring area. Firstly, wavelet de-noising method was used to reduce the noise of original signals, and eliminate the effect of noise on fractal dimensions. Secondly, the fractal dimensions based on fractal theory were got from the de-noised signals, including box dimension, information dimension, and correlation dimension. Finally, the relationship between the fractal dimensions and tool wear states was studied; the affinities between the known and unknown states can be obtained through fuzzy c-mean clustering algorithm; tool wear states can be recognized by those affinities based on fractal dimensions. The experiment results demonstrate that wavelet de-noising method can efficiently eliminate the effect of noise on fractal dimensions, and tool wear states can be real-timely and accurately recognized through the fuzzy clustering analysis on fractal dimensions. DOI: http://dx.doi.org/10.11591/telkomnika.v11i1.1887]]>
