Online prediction of surface roughness in peripheral milling processes

An online surface roughness prediction module for peripheral end milling in High Speed Machining was developed. An Artificial Neural Network framework integrated five cutting parameters and one process variable signal. Vibration signal in the workpiece showed high correlation with the surface roughness. This signal was pre-processed as Mel Frequency Cesptrum Coefficients. This could be a practical solution for a wide cutting conditions with several Aluminium alloys and cutting tools. Results were validated by using an industrial High Speed Machining center.