On a simulation-based chatter prediction system by integrating relative entropy and dynamic cutting force

Abstract

This paper aims to develop a simulation-based chatter prediction system using relative entropy—Kullback-Leibler divergence (KLD), and the NC program can be modified to become non-chatter. Chatter is one of the major concerns when machining mechanical components on a CNC machine. In general, the majority of the previous research methods achieved non-chatter stable machining by assigning the appropriate machining parameters: (1) spindle speed, (2) feed rate, and (3) depth of cut based on the generated SLD (stability lobe diagram). Non-chatter stable machining can also be accomplished by manually adjusting the spindle override percentage on the operation panel or the values in the CNC controller via networking once chatter is detected during the machining processes. The creation of SLD must consider two essential parameters: cutting force coefficients (CFCs) and frequency response function (FRF). The CFCs can be obtained from cutting experiment data related to a paired tool and workpiece, and the FRF can be calculated from the tapping test experiment. Then, the CFCs and FRF are stored in the database of the developed system. The simulation-based chatter prediction calculates the KLD value based on the relativity of the dynamic cutting force and the static cutting forces so as to predict whether there is chatter in the NC program or not. The NC program can be adjusted to become non-chatter if there is chatter predicted. The proposed method has been successfully verified through on-site machining, showing very promising achievement.