Investigation on milling chatter identification at early stage with variance ratio and Hilbert–Huang transform

Chatter occurs as an unexpected and unstable phenomenon in milling process, imposing an extremely negative effect on the workpiece and machining system. Therefore, a method that can identify chatter at an early stage is desperately needed. However, some aspects in terms of effectiveness, robustness, and practicality with existing methods deserve further improvement. In this paper, the characteristics of chatter in different stages are investigated. Considering the properties of a vibration signal when the onset of chatter occurs, an adaptive filter is designed to remove the spindle speed-related frequency components, and the chatter-related components can be amplified simultaneously with the filter. Next, the variance ratio (VR) of the filtered signal series to the original signal series is defined as the chatter indicator, which is very sensitive to chatter. After chatter is detected, a time frequency analysis method based on ensemble empirical mode decomposition (EEMD) and the Hilbert–Huang transform (HHT) is introduced to estimate the dominant chatter frequency. Milling experiments with different configurations of cutting conditions are performed and the results show that all the chatter can be detected at an early stage. In addition, the transients of the vibration signal caused by discontinuity of workpiece geometry or inhomogeneous material can be distinguished from the chatter.