Novel on-line monitoring method of cutting chatter based on neural network pattern recognition technique

A novel cutting chatter monitoring method is presented by investigating the characteristics of probability density of the cutting system vibration response. When the cutting is stable, the vibration response can be regarded as random noise and the corresponding probability density function is normal distributed. However, sinusoidal components dominate the vibration response and its probability density function is U-shaped distributed when the cutting process experiences chatter. Thus the probability density of the vibration response is sensitive to cutting chatter and is employed as the monitoring parameter in this study. Neural network pattern recognition technique is introduced to recognize cutting chatter considering its excellent capability of classification. Experiments conducted on an engine lathe confirm that the probability density of the vibration response a suitable monitoring parameter and cutting chatter can be successfully detected by the proposed method.