Study on Identification Method of Tool Wear Based on Singular Spectrum Analysis and Support Vector Machine

摘要

Acoustic Emission signal reflecting the tool wear state is made by phase space reconstruction that uses mutual information method and Cao method to determine time delay and embedding dimension for constructing phase space matrix. After reconstruction, by calculating singular spectral of phase space matrix, based on which characteristic vector is constructed. These characteristic vectors are combined with the Support Vector Machine for training, which supports Support Vector Machine classifier model to predict new data. Compared with classifier model gotten by AE signal being directly put into Support Vector Machine after phase space reconstruction, the AE signals based on KC9125 tool cutting 40CrNiMoA can increase forecasting accuracy from 90% to raise 98%.