Extenuating Chatter Vibration in Milling Process Using a New Ensemble Approach

Background Milling process contributes 60-70% of the finishing process in machining industries. Surface finish of the product is greatly influenced by the nature of chatter vibration initiated during milling due to its self-generating chatter phenomenon. Purpose To ascertain the stability regimes in milling, in this work, a novel methodology based on Spline-Based Local Mean Decomposition (SB-LMD) and Artificial Neural Network (ANN) is proposed. Method For this purpose, experimentally acquired audio signals in milling operation have been processed using a SB-LMD technique to extract tool chatter features. Furthermore, three ANN training algorithms viz. Resilient Propagation (RP), Conjugate Gradient Based (CGP) and Levenberg-Marquardt Algorithm (LM) and two activation functions viz. Hyperbolic Tangent Sigmoid Transfer Function (TANSIG) and Log Sigmoid Transfer Function (LOGSIG) has been used to train the data set. Among these training algorithms and activation functions, most suitable combination has been selected and further invoked to develop prediction model of chatter severity in terms of Chatter Index (CI). Results Results showed that the proposed methodology is quite suitable for ascertaining the stable milling parameters that will result in higher productivity along with better surface finish. Conclusion A technique to extract chatter frequency corresponding to the tool chatter has been developed and tested. Moreover, TANSIG with optimal neurons in hidden layer is found to be the most suitable one for the prediction of CI with an average deviation of 3.11%.

Automated summary

A novel methodology based on Spline-Based Local Mean Decomposition (SB-LMD) and Artificial Neural Network (ANN) is proposed to ascertain the stability regimes in milling. Audio signals acquired during milling operation are processed using SB-LMD technique to extract tool chatter features. Different training algorithms and activation functions are used to train the data set, and the most suitable combination is selected to develop a prediction model of chatter severity. The results show that the proposed methodology is effective in predicting chatter severity and improving productivity and surface finish in milling.