关键词: 刀具磨损, 振动异常, 主轴驱动电流, 卷积神经网络, 电流杂波

Abstract: To ensure tool life and control the rejects rate of work-piece,a method to extract clutter induced by tool wear and abnormal vibration in spindle current was proposed.In addition,wear identification of end milling cutter was realized by Convolutional Neural Network (CNN).Based on the results of irregular clutter components in spindle current caused by tool wear and abnormal vibration,Fourier series fitting was used to decompose current into harmonic components and current clutter.Harmonic components reflected the quasi-static change in the current RMS and current clutter reflected the cutting edge and flank wear status of the end milling cutter and abnormal vibration.The current clutter signals were input into CNN for feature extraction and classification of end milling cutter status.Experiments results showed that the proposed method could eliminate the influence of cutting vibration and parameters on the accuracy of tool wear status identification,realize the accurate status identification of the end milling cutter and lay the foundation for predicting the remaining life of the end milling cutter and formulating the end milling cutter changing rules scientifically and rationally under complex conditions.

Key words: tool wear, abnormal vibration, spindle current, convolutional neural network, current clutter