Feature Extraction of Turing Tool Wear Based on J-EEMD

Abstract

In the monitoring of cutting tool state, a large number of redundant information is contained in the sensor signal. Therefore, it is obviously not conducive to pattern recognition, and difficult to classify the tool wear state correctly from the available sensors. The test platform that had real-time information collection of the vibration and acoustic emission signals in turning was built. Observed signals were adaptively processed using the method of ensemble empirical mode decomposition introduced joint approximate diagonalization of eigenmatrices (J-EEMD). This method is based on the characteristics of the signal itself decomposed into several intrinsic mode functions (IMF), and then transforms the energy ratio between the IMF. The white noise of each IMF component has been eliminated by introducing JADE algorithm during the signal decomposition. Compared with the EEMD algorithm, the decomposition efficiency is significantly improved. The experiments showed that the method could identify the different states of tool wear, if applied to feature extraction of vibration and acoustic emission signal in the cutting process.

 DOI: http://dx.doi.org/10.11591/telkomnika.v11i10.3423