Diagnosis Analysis Of Chatter And Parameter Optimization In Milling

Milling is a very important way of processing and manufacturing,which is widely used in the machinery manufacturing and processing industry.In the process of machining thin-walled workpiece,because of the relatively low stiffness and stability of the workpiece,it is easy to lead to cutting chatter.The occurrence of chatter will limit the machining efficiency,affect the surface quality of machining,and shorten the service life of tools.Therefore,how to realize the target which high efficiency,high precision and high reliability of milling has become a difficult problem to be solved in the manufacturing industry.On the basis of summarizing the research status in related fields,the chatter problem in milling process is studied in this paper.The milling force model is established.Based on the theory of machining dynamics,the relative transfer function is introduced to establish the prediction model of milling stability;The milling force signal is collected and the signal processing method is used to extract the physical characteristics of the signal to diagnose the chatter;Based on the above chatter analysis,a chatterless multi-objective model is established to optimize the milling parameters.Content of the study is presented as follow:(1)Using the infinitesimal integral method established the flat-head milling cutter model,and then the mechanism of chatter vibration in milling process was studied.The relative transfer function of the processing system was formed by considering the relative transfer function between the cutter and the workpiece,and as a basis for the stability of the study.The stability curve of milling chatter is drawn by simulation,and its effectiveness is verified by experiments.(2)The ensemble empirical mode decomposition(EMD)is introduced and its effectiveness is verified by simulation signals.This method is introduced into the feature extraction of chatter signals,and the representative sample entropy and energy of different IMF components are extracted to form feature vectors.Finally,the principal component analysis method is used to reduce the dimension of the feature vector,which is ready for the subsequent chatter diagnosis and identification.(3)The vibration feature vector extracted,trained using support vector machine learning ability,and to test the effect of different types of kernel function,grid search and genetic algorithm of penalty factor c and g kernel parameter optimization,comparison of different optimization results,finally obtained good classification effect.(4)The orthogonal experiment of titanium alloy thin walled parts was carried out,and the prediction model of surface roughness was obtained.With the maximum material removal rate and surface roughness as the optimization objective,the chatter stability limit surface,feed speed and surface roughness are set as constraint conditions.Three different optimization models are established respectively,and the particle swarm optimization algorithm is used to solve the optimal solution of each model,and the model is compared and analyzed.