Application of linear regression model on chatter threshold delineation

Abstract

This research is to parameterize the variation in reconstructed phase space attractor Poincaré sections during the milling process in order to identify the transition of system from stable to unstable condition. Vibration is continuously acquired during the milling process and converted to the form of reconstructed time series. A regression model is developed to compute the trend of changes in the reconstructed phase space attractor Poincaré sections and output a numerical value indicating the level of stability. A threshold value of the trend value defines the boundary of stable and unstable states of milling process. The method monitors the state of system without the necessity to have analytical relationship between cutting parameters and milling dynamics. It can also be used as an online monitoring process to detect the onset of chatter.