Study on identification method of tool wear based on feature fusion and least squares support vector machine

For accurately identifying the condition of the tool wear in vector machines, a novel feature vector extraction methods based on fusing wavelet packet multi-scale information entropy (Frequency domain)and AR model coefficients(Time Domain) of Acoustic emission signal of tool wear is proposed. In order to reduce the dimension of feature vector, the analysis method of kernel principal component analysis method is adopted. The new feature vector is put into lease squares support vector machine to train and identify the tool wear state. The identification results proved that the method using feature fusion obtain higher recognition rate than that using the Single feature.