Enhancement of the low-level components of milling vibration signals by stochastic resonance

The enhancement of vibration signals by the addition of a random signal component was investigated. Vibration measurements during end-milling were made using an accelerometer mounted on the workpiece. A large number of tests were run with varying spindle speed, table speed, axial depth of cut, radial depth of cut and tool condition. Vibration signal frequency spectra were dominated by the cutting edge passing frequency, and so individual cutting edge effects such as cutting edge chipping or differences in amount of edge wear could not be identified easily. Similarly, machine characteristics such as spindle runout were overshadowed in vibration spectra by the dominant cutting edge passing frequency. To enhance the low-level signal components, a signal analysis procedure was constructed in which random components were added to accelerometer signals, a threshold signal level set and frequency spectra calculated from signal components exceeding the threshold. Enhancement of spectra peaks associated with low-level signal components was demonstrated. The effects of measurement system threshold level and added noise level on signal enhancement were determined.