A study on real-time detecting of machine tool chatter

A unified nonlinear time series analysis approach to the problem of feature extraction of machine tool chatter is developed. Considering the mechanism of chatter development, the procedure proposes the limit cycle behavior of self-excited random vibration to be the intrinsic index of chatter occurrence, and provides the exponential autoregressive (EAR) model to extract the index from on-line measured machining signal. In order to real-time implement: this procedure, a quickly estimating method of the EAR model is also provided by introducing a heuristic determination of the nonlinear coefficient in the model. Finally both the simulation and experimental data are used to shown the effectiveness of the procedure.