Identification of blunt-process at wood-milling tools by time-frequency analysis

This paper deals with the time-frequency-distribution (TFD) of transient nonstationary vibrations, which appears at the milling of wood or wood-like materials (chipboards) by means of wood processing tools. Correlating with the wear of cutting edges features can be extracted out of the measured acceleration signals by means of time-frequency analysis. We developed two different algorithms for recognition of the actual state of wear. Each of the cutting edges of a multiblade milling head can be attached to a certain class, which describes the degree of wear. All classes are defined by the accompanying vector of characteristic features in the multidimensional feature-space. For pattern recognition we use a data-based classifier. This approach is furthermore demonstrated by an example.