The Diagnosis of Tool Wear Based on EMD and GA-B-Spline Network

In view of the strong background noise involved in vibration signal of tool wear and the difficulty to obtain fault frequencies, this paper proposed a tool wear fault feature extraction method based on morphological filters-singularity value decomposition (SVD) with empirical mode decomposition (EMD). Firstly, an experiment system of the cutting tool wear monitoring was set up and a variety of data coming from vibratory sensor were collected, then, the pulse components from the original signal were inhibited by morphological filters and the signal sequences removed outlier were reconstructed, the attractor track matrix was decomposed using SVD for further noise reduction, and then we got weak signal failure frequency after the de-noise signals were decomposed with EMD. Finally, tool wear was identified by GA-B-spline neural network. B-spline networks were trained using genetic algorithms to search for global optimization. The experimental results shown that the diagnosis approach put forward in this paper could identify tool wear fault patterns effectively in noise background. [PUBLICATION ABSTRACT]