Effective monitoring of machining operations, for tool breakage and wear in automotive powertrain manufacturing, has proven difficult despite significant activity in industry and universities. Various monitoring schemes have been tried with different sensing strategies and algorithms. Despite these efforts, results have been mixed, with systems working in some operations and not in others or systems requiring too much attention on the plant floor. All this, while the need for reliable monitors, due to emphasis on productivity and quality, continues to increase. Some of the reasons for the limited success of monitoring systems include changing machine condition, variability of cutting edge on tools due to regrind, the variability of incoming part material together with different possible failure modes for the tools. The shortcoming of existing monitoring systems has been even more apparent in multi-spindle and spindle cluster stations whereby a number of tools operate simultaneously, therefore adding to the complexity. The frequency and time domain vibration1 based monitoring strategy described in this paper for breakage detection addresses some of the shortcomings of the available systems by concentrating on discriminants which produce high signal to noise ratio. Also, by recognizing that the tool failure signatures may vary on different machines while performing the same machining operations with similar tools. This can be attributed to different structural characteristics of the spindles and other elements of the machine tools. Therefore, in this paper for the first time, a pre-characterization2 scheme is proposed to capture this vital information and to utilize it for adding further robustness to tool monitoring.

This content is only available via PDF.
You do not currently have access to this content.