Cutting chatter feature extraction method based on mean square frequency and Empirical Mode Decomposition

Chattering has a great influence on the stability of machine tool processing, which seriously restricts the result & efficiency of machine tool processing. In order to realize online monitoring and prediction of cutting chatter, a method for extracting chatter characteristics based on mean square frequency and empirical mode decomposition is proposed. Through frequency domain analysis of the cutting force signal, the mean square frequency ratio coefficient of the wavelet packet flutter characteristic frequency band is calculated as the characteristic T₁; By calculating the correlation between the IMF component and the original signal, the IMF with strong correlation is selected and the energy entropy is calculated as the characteristic T₂. The research results show that both methods can effectively identify chatter. The overall rate of change of T₁ is small, reaching 20%, but the early warning of flutter can be achieved faster. T₂ is slightly inadequate in early warning, but its overall rate of change is relatively large, close to 80%, which can achieve accurate alarms, form an effective complement to T₁, and achieve pre-alarms for cutting chatter.