In Situ Chatter Frequency Prediction Using Torque Data From a Wireless Sensor Integrated Tool Holder

A method is demonstrated to estimate chatter frequencies in real time from cutting torque data using formant frequency tracking. Formant frequencies are derived from the torque data using linear predictive coding (LPC) methods, similar to algorithms used in speech recognition. The estimated frequency response is observed to change throughout a cut as a function of both engagement and cut geometry. Torque data is collected at high bandwidth from a wireless sensor integrated end milling tool holder. The technique is found to be effective and repeatable for forecasting regenerative chatter frequencies in real time. Chatter frequencies predicted during non-chatter conditions correctly estimate the actual chatter condition. To demonstrate the technique, a number of experimental cuts are conducted and discussed.