Research On Comprehensive On-line Prediction Of Regenerative Cutting Chatter

Cutting chatter is an intensive and relative vibration between the cutting tool and work piecein process. It is one of the important problem because it decreases the quality of products andthe efficiency of processing, shortens life expectancies of cutting tools and machines. So, it is ofsignificant importance for accelerating the development of manufacturing to analyze andcontrol cutting chatter.Firstly, several models of cutting chatter mechanism were researched and the cutting chatterresearch achievement of on-line monitoring, prediction and control was summarized. Based onthe regenerative chatter system, dynamic model and equation of the regenerative chatter systemwith the time-delay feedback was researched. The theoretical analysis of its characteristics atports both time and frequency domain varies provided foundation for simulations.Secondly, based on the chatting characteristics at ports both timeand frequency domain varies, the signal analysis theory of chatting was simulated. In the timedomain, using the variance theory, chatting signal stability, transition and flutter state wasresearched, and the main frequency and autocorrelation function relationship was simulatedusing the autocorrelation theory. The resolution of autocorrelation and Burg algorithm and thepower spectrum estimation changes result from Chatter signal frequency were researchedthrough parameterization power spectral estimation.Thirdly, the online chatting signal acquisition and analysis system using acceleration sensorsand FFT analyzer was formatted, and it online-monitored cutting process on C646Shenyangmachine and collected cutting state data. The sign characteristics at ports bothtime and frequency domain was analyzed using chatting signal analysis theory. Thecharacteristic quantity reflecting the chatter time domain and frequency domain characteristicswas found, and its feasibility, accuracy and timeliness was analyzed, for construction of thecutting chatter comprehensive prediction function do basic.Finally, the cutting chatter comprehensive prediction function was constructed, using thevariance ratioB reflecting the cutting chatter time domain features and the autocorrelationfunction reflecting the frequency domain characteristics. The chatting threshold ofcomprehensive prediction function was found, based on experimental data and previous experiments. Chatting monitoring and forecasting system was established according tocomprehensive prediction function, and its ability to predict cutting chatter was simulated byactual chatter signals, and the simulation result proved that the chatting monitoring andforecasting system can accurately predict the occurrence of cutting chatter, and avoidmisjudgment due to chance.