Analysis of acoustic signals due to tool wear during machining.

摘要

In response to frequent variations in market demand, reconfigurable manufacturing systems (RMS) are being developed to adapt production systems to capacity and technological changes. In this research, a methodology for analyzing sound based tool wear monitoring was addressed to facilitate a reduction in ramp-up time for RMS. A model describing the relationship between tool wear and sound generation was developed and analyzed assuming the vibration of the tool holder-insert combination as the main sources of sound associated with tool wear. Interaction between asperities on the workpiece and flank surface of the cutting tool is considered as a source of system excitation. In modeling the interaction on the flank surface, the asperities on the surfaces are represented as a trapezoidal series function with normal distribution. The stiffness between the interacting surfaces is assumed to be proportional to flank wear. The tool holder-insert combination is simplified as a cantilever beam excited by external forces including cutting force, thrust force, friction force, and the excitation force generated from the asperity interaction. Based on this model, the sound-based monitoring performance that is associated with the process characteristics can be estimated for different processes. As a result of these estimations, the sensor selection and signal processing design could be enhanced. Furthermore, a filter design was introduced to reduce the noise effect. A new method for reducing background noise while monitoring tool wear during machining is introduced by combining spectral subtraction and a Wiener filter. In addition, the optimum bandwidth and feature dimensions, as well as the sensor types for tool wear monitoring were analyzed. An index J based on the class-mean scatter feature selection criterion provides a more quantitative basis for evaluating the system characteristics.