ON-LINE IDENTIFICATION AND CONTROL OF MACHINING CHATTER IN TURNING THROUGH DYNAMIC DATA SYSTEM METHODOLOGY

The time-varying stability of the machining process necessitates a technique of on-line chatter identification and control. In a six-stage study, such a technique was sought. (1) Mathematical models of the machining process with an inherent stochastic nature were developed as discrete ARMA (n,n-1) models. Based on off-line analysis, the peak of power spectral density corresponding to the dynamic mode of workpiece fundamental natural frequency serves as a simple and reliable index of stability for on-line chatter identification. (2) The workpiece fundamental natural frequency of a given CNC lathe was identified using random excitation and impulse response methods. A band pass filter in the range of 70-170 Hz served to isolate the vibration signal. The bivariate ARMA modeling technique was employed for modal analysis of the chuck-workpiece-tailstock system. (3) A self-learning scheme to determine the dynamic stability limit, based on the quality control chart technique, has been established so that the threshold stability limit of each cutting process need not be initialized before cutting. (4) A simple and fast linearized adaptive AR modeling technique was adopted for on-line machining process identification. An AR model of order six was determined as an adequate linearized approximation of the machining process. (5) The strategy of changing speed and feed incrementally to find stable cutting conditions without sacrificing productivity was proposed. (6) A Chatter Suppression Controller was designed and interfaced to the PT15CNC lathe. Cutting tests demonstrated the successful use of the above strategies.;The theoretical derivation and developed strategy presented in this study have provided a solid basis for developing a chatter free lathe.