Robotic Milling Chatter Identification Based on the Convolutional Neural Network
Serial industrial robots have the characteristics of low stiffness and are prone to chatter during milling,reducing the surface quality and dimensional accuracy of the workpiece.To solve the problem of robotic milling chatter identification,a convolutional neural network based chatter identification method is proposed.Robotic milling experiments were carried out to collect the vibration acceleration signals under different machining conditions,and the power spectrum entropy difference and root mean square of the sig-nals were extracted as the characteristic matrix.Then two dimensional scatter image data sets representing different milling states of the robotic milling were constructed by the characteristic matrix.The convolution-al neural network algorithm is used to construct the intelligent chatter identification model of robotic milling.In order to verify the effectiveness of the proposed method,comparison analysis and verification are carried out with existing methods.The results show that the robotic chatter identification model based on convolution-al neural network can effectively identify different states in the robotic milling process,and its recognition ac-curacy can reaches 95.10％,which is superior to the model constructed by the support vector machine.
万方数据
维普
